Compare commits
2 Commits
d424d56143
...
572e1d5abd
| Author | SHA1 | Date | |
|---|---|---|---|
| 572e1d5abd | |||
| 504f0d7935 |
24
index.html
24
index.html
@ -3,9 +3,9 @@
|
||||
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
|
||||
<html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="en">
|
||||
<head>
|
||||
<!-- 2021-09-01 mer. 17:04 -->
|
||||
<!-- 2021-09-02 jeu. 10:02 -->
|
||||
<meta http-equiv="Content-Type" content="text/html;charset=utf-8" />
|
||||
<title>Complementary Filters Shaping Using $\mathcal{H}_\infty$ Synthesis</title>
|
||||
<title>A new method of designing complementary filters for sensor fusion using the $\mathcal{H}_\infty$ synthesis</title>
|
||||
<meta name="author" content="Thomas Dehaeze" />
|
||||
<meta name="generator" content="Org Mode" />
|
||||
<link rel="stylesheet" type="text/css" href="https://research.tdehaeze.xyz/css/style.css"/>
|
||||
@ -34,7 +34,7 @@
|
||||
|
|
||||
<a accesskey="H" href="../index.html"> HOME </a>
|
||||
</div><div id="content" class="content">
|
||||
<h1 class="title">Complementary Filters Shaping Using \(\mathcal{H}_\infty\) Synthesis
|
||||
<h1 class="title">A new method of designing complementary filters for sensor fusion using the \(\mathcal{H}_\infty\) synthesis
|
||||
<br />
|
||||
<span class="subtitle">Dehaeze Thomas, Verma Mohit and Collette Christophe</span>
|
||||
</h1>
|
||||
@ -55,9 +55,9 @@ Such synthesis method is further extended for the shaping of more than two compl
|
||||
</p>
|
||||
</blockquote>
|
||||
|
||||
<div id="outline-container-org1d8b653" class="outline-2">
|
||||
<h2 id="org1d8b653">Journal Article (<a href="journal/dehaeze21_desig_compl_filte.pdf">link</a>)</h2>
|
||||
<div class="outline-text-2" id="text-org1d8b653">
|
||||
<div id="outline-container-org8d13d5d" class="outline-2">
|
||||
<h2 id="org8d13d5d">Journal Article (<a href="journal/dehaeze21_desig_compl_filte.pdf">link</a>)</h2>
|
||||
<div class="outline-text-2" id="text-org8d13d5d">
|
||||
<p>
|
||||
The paper has been created <a href="https://orgmode.org/">Org Mode</a> (generating <a href="https://www.latex-project.org/">LaTeX</a> code) under <a href="https://www.gnu.org/software/emacs/">Emacs</a>.
|
||||
</p>
|
||||
@ -88,9 +88,9 @@ Dehaeze, T., Vermat, M., & Collette, C., A new method of designing complemen
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org6ba84bc" class="outline-2">
|
||||
<h2 id="org6ba84bc">Matlab Scripts (<a href="matlab/dehaeze21_desig_compl_filte_matlab.html">link</a>)</h2>
|
||||
<div class="outline-text-2" id="text-org6ba84bc">
|
||||
<div id="outline-container-org71c1b40" class="outline-2">
|
||||
<h2 id="org71c1b40">Matlab Scripts (<a href="matlab/dehaeze21_desig_compl_filte_matlab.html">link</a>)</h2>
|
||||
<div class="outline-text-2" id="text-org71c1b40">
|
||||
<p>
|
||||
All the <a href="https://fr.mathworks.com/">Matlab</a> code that was used for the paper are accessible so that all the results are reproducible.
|
||||
The source code is hosted on <a href="https://zenodo.org/record/5361920">Zenodo</a>.
|
||||
@ -123,9 +123,9 @@ Dehaeze Thomas. (2021). A New Method of Designing Complementary Filters for Sens
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org827f683" class="outline-2">
|
||||
<h2 id="org827f683">Tikz Figures (<a href="tikz/dehaeze21_desig_compl_filte_tikz.html">link</a>)</h2>
|
||||
<div class="outline-text-2" id="text-org827f683">
|
||||
<div id="outline-container-org1b73cc9" class="outline-2">
|
||||
<h2 id="org1b73cc9">Tikz Figures (<a href="tikz/dehaeze21_desig_compl_filte_tikz.html">link</a>)</h2>
|
||||
<div class="outline-text-2" id="text-org1b73cc9">
|
||||
<p>
|
||||
All the figures for the paper have been generated using <a href="https://sourceforge.net/projects/pgf/">TikZ</a>.
|
||||
</p>
|
||||
|
||||
@ -1,4 +1,4 @@
|
||||
#+TITLE: Complementary Filters Shaping Using $\mathcal{H}_\infty$ Synthesis
|
||||
#+TITLE: A new method of designing complementary filters for sensor fusion using the $\mathcal{H}_\infty$ synthesis
|
||||
:DRAWER:
|
||||
#+SUBTITLE: Dehaeze Thomas, Verma Mohit and Collette Christophe
|
||||
|
||||
|
||||
@ -1,4 +1,4 @@
|
||||
#+TITLE: A new method of designing complementary filters for sensor fusion using the $\mathcal{H}_\infty$ synthesis
|
||||
#+TITLE: Designing complementary filters for sensor fusion using $\mathcal{H}_\infty$ synthesis
|
||||
:DRAWER:
|
||||
#+LATEX_CLASS: elsarticle
|
||||
#+LATEX_CLASS_OPTIONS: [preprint, sort&compress]
|
||||
@ -13,7 +13,7 @@
|
||||
#+LATEX_HEADER_EXTRA: \cortext[cor1]{Corresponding author. Email Address: thomas.dehaeze@esrf.fr}
|
||||
|
||||
#+LATEX_HEADER_EXTRA: \address[a1]{European Synchrotron Radiation Facility, Grenoble, France}
|
||||
#+LATEX_HEADER_EXTRA: \address[a2]{University of Li\`{e}ge, Department of Aerospace and Mechanical Engineering, 4000 Li\`{e}ge, Belgium.}
|
||||
#+LATEX_HEADER_EXTRA: \address[a2]{University of Li\`{e}ge, PML, Department of Aerospace and Mechanical Engineering, 4000 Li\`{e}ge, Belgium.}
|
||||
#+LATEX_HEADER_EXTRA: \address[a3]{CSIR --- Structural Engineering Research Centre, Taramani, Chennai --- 600113, India.}
|
||||
#+LATEX_HEADER_EXTRA: \address[a4]{Universit\'{e} Libre de Bruxelles, Precision Mechatronics Laboratory, BEAMS Department, 1050 Brussels, Belgium.}
|
||||
|
||||
@ -41,7 +41,16 @@
|
||||
("\\paragraph{%s}" . "\\paragraph*{%s}")
|
||||
("\\subparagraph{%s}" . "\\subparagraph*{%s}"))
|
||||
)
|
||||
;; Remove automatic org headings
|
||||
(defun my-latex-filter-removeOrgAutoLabels (text backend info)
|
||||
"Org-mode automatically generates labels for headings despite explicit use of `#+LABEL`. This filter forcibly removes all automatically generated org-labels in headings."
|
||||
(when (org-export-derived-backend-p backend 'latex)
|
||||
(replace-regexp-in-string "\\\\label{sec:org[a-f0-9]+}\n" "" text)))
|
||||
|
||||
(add-to-list 'org-export-filter-headline-functions
|
||||
'my-latex-filter-removeOrgAutoLabels)
|
||||
|
||||
;; Automatic delete org org-comments
|
||||
(defun delete-org-comments (backend)
|
||||
(loop for comment in (reverse (org-element-map (org-element-parse-buffer)
|
||||
'comment 'identity))
|
||||
@ -84,57 +93,57 @@ Sensor fusion \sep{} Complementary filters \sep{} $\mathcal{H}_\infty$ synthesis
|
||||
** Introduction to Sensor Fusion :ignore:
|
||||
|
||||
Measuring a physical quantity using sensors is always subject to several limitations.
|
||||
First, the accuracy of the measurement will be affected by several noise sources, such as the electrical noise of the conditioning electronics being used.
|
||||
First, the accuracy of the measurement will be affected by several noise sources, such as electrical noise of the conditioning electronics being used.
|
||||
Second, the frequency range in which the measurement is relevant is bounded by the bandwidth of the sensor.
|
||||
One way to overcome these limitations is to combine several sensors using a technique called "sensor fusion"\nbsp{}cite:bendat57_optim_filter_indep_measur_two.
|
||||
Fortunately, a wide variety of sensors exist, each with different characteristics.
|
||||
By carefully choosing the fused sensors, a so called "super sensor" is obtain that combines benefits of individual sensors and yields a better estimate of the measured physical quantity. \par
|
||||
By carefully choosing the fused sensors, a so called "super sensor" is obtained that combines benefits of individual sensors. \par
|
||||
|
||||
** Advantages of Sensor Fusion :ignore:
|
||||
|
||||
In some situations, sensor fusion is used to increase the bandwidth of the measurement\nbsp{}cite:shaw90_bandw_enhan_posit_measur_using_measur_accel,zimmermann92_high_bandw_orien_measur_contr,min15_compl_filter_desig_angle_estim.
|
||||
For instance, in\nbsp{}cite:shaw90_bandw_enhan_posit_measur_using_measur_accel, the bandwidth of a position sensor is increased by fusing it with an accelerometer providing the high frequency motion information.
|
||||
For other applications, sensor fusion is used to obtain a estimate of the measured quantity with lower noise\nbsp{}cite:hua05_low_ligo,hua04_polyp_fir_compl_filter_contr_system,plummer06_optim_compl_filter_their_applic_motion_measur,robert12_introd_random_signal_applied_kalman.
|
||||
More recently, the fusion of sensors measuring different physical quantities has been proposed to obtained interesting properties for control\nbsp{}cite:collette15_sensor_fusion_method_high_perfor,yong16_high_speed_vertic_posit_stage.
|
||||
In\nbsp{}cite:collette15_sensor_fusion_method_high_perfor, an inertial sensor used for active vibration isolation is fused with a sensor collocated with the actuator to improve the stability margins of the feedback controller. \par
|
||||
For other applications, sensor fusion is used to obtain an estimate of the measured quantity with lower noise\nbsp{}cite:hua05_low_ligo,hua04_polyp_fir_compl_filter_contr_system,plummer06_optim_compl_filter_their_applic_motion_measur,robert12_introd_random_signal_applied_kalman.
|
||||
More recently, the fusion of sensors measuring different physical quantities has been proposed to obtain interesting properties for control\nbsp{}cite:collette15_sensor_fusion_method_high_perfor,yong16_high_speed_vertic_posit_stage.
|
||||
In\nbsp{}cite:collette15_sensor_fusion_method_high_perfor, an inertial sensor used for active vibration isolation is fused with a sensor collocated with the actuator for improving the stability margins of the feedback controller. \par
|
||||
|
||||
** Applications :ignore:
|
||||
|
||||
Practical applications of sensor fusion are numerous.
|
||||
It is widely used for the attitude estimation of several autonomous vehicles such as unmanned aerial vehicle\nbsp{}cite:baerveldt97_low_cost_low_weigh_attit,corke04_inert_visual_sensin_system_small_auton_helic,jensen13_basic_uas and underwater vehicles\nbsp{}cite:pascoal99_navig_system_desig_using_time,batista10_optim_posit_veloc_navig_filter_auton_vehic.
|
||||
Naturally, it is of great benefits for high performance positioning control as shown in\nbsp{}cite:shaw90_bandw_enhan_posit_measur_using_measur_accel,zimmermann92_high_bandw_orien_measur_contr,min15_compl_filter_desig_angle_estim,yong16_high_speed_vertic_posit_stage.
|
||||
Sensor fusion was also shown to be a key technology to improve the performances of active vibration isolation systems\nbsp{}cite:tjepkema12_sensor_fusion_activ_vibrat_isolat_precis_equip.
|
||||
This is particularly apparent for the isolation stages of gravitational wave observer\nbsp{}cite:collette15_sensor_fusion_method_high_perfor,heijningen18_low such as the ones used at the LIGO\nbsp{}cite:hua05_low_ligo,hua04_polyp_fir_compl_filter_contr_system and at the VIRGO\nbsp{}cite:lucia18_low_frequen_optim_perfor_advan. \par
|
||||
Sensor fusion was also shown to be a key technology to improve the performance of active vibration isolation systems\nbsp{}cite:tjepkema12_sensor_fusion_activ_vibrat_isolat_precis_equip.
|
||||
Emblematic examples are the isolation stages of gravitational wave detectors\nbsp{}cite:collette15_sensor_fusion_method_high_perfor,heijningen18_low such as the ones used at the LIGO\nbsp{}cite:hua05_low_ligo,hua04_polyp_fir_compl_filter_contr_system and at the VIRGO\nbsp{}cite:lucia18_low_frequen_optim_perfor_advan. \par
|
||||
|
||||
** Kalman Filtering / Complementary filters :ignore:
|
||||
|
||||
There are mainly two ways to perform sensor fusion: either using a set of complementary filters\nbsp{}cite:anderson53_instr_approac_system_steer_comput or using Kalman filtering\nbsp{}cite:brown72_integ_navig_system_kalman_filter,odry18_kalman_filter_mobil_robot_attit_estim.
|
||||
For sensor fusion applications, both methods are sharing many relationships\nbsp{}cite:brown72_integ_navig_system_kalman_filter,higgins75_compar_compl_kalman_filter,robert12_introd_random_signal_applied_kalman,becker15_compl_filter_desig_three_frequen_bands.
|
||||
However, for Kalman filtering, we are forced to make assumption about the probabilistic character of the sensor noises\nbsp{}cite:robert12_introd_random_signal_applied_kalman whereas it is not the case for complementary filters.
|
||||
Furthermore, the main advantages of complementary filters over Kalman filtering for sensor fusion are their very general applicability, their low computational cost\nbsp{}cite:higgins75_compar_compl_kalman_filter, and the fact that they are very intuitive as their effects can be easily interpreted in the frequency domain. \par
|
||||
However, for Kalman filtering, we are forced to make assumption about the probabilistic character of the sensor noises\nbsp{}cite:robert12_introd_random_signal_applied_kalman whereas it is not the case with complementary filters.
|
||||
Furthermore, the advantages of complementary filters over Kalman filtering for sensor fusion are their general applicability, their low computational cost\nbsp{}cite:higgins75_compar_compl_kalman_filter, and the fact that they are intuitive as their effects can be easily interpreted in the frequency domain. \par
|
||||
|
||||
** Design Methods of Complementary filters :ignore:
|
||||
|
||||
A set of filters is said to be complementary if the sum of their transfer functions is equal to one at all frequencies.
|
||||
For the earliest used of complementary filtering for sensor fusion, analog circuits were used to physically realize the filters\nbsp{}cite:anderson53_instr_approac_system_steer_comput.
|
||||
Analog complementary filters are still used today\nbsp{}cite:yong16_high_speed_vertic_posit_stage,moore19_capac_instr_sensor_fusion_high_bandw_nanop, but most of the time they are now implemented numerically as it allows for much more flexibility. \par
|
||||
In the early days of complementary filtering, analog circuits were employed to physically realize the filters\nbsp{}cite:anderson53_instr_approac_system_steer_comput.
|
||||
Analog complementary filters are still used today\nbsp{}cite:yong16_high_speed_vertic_posit_stage,moore19_capac_instr_sensor_fusion_high_bandw_nanop, but most of the time they are now implemented digitally as it allows for much more flexibility. \par
|
||||
|
||||
Several design methods have been developed over the years to optimize complementary filters.
|
||||
The easiest way to design complementary filters is to use analytical formulas.
|
||||
Depending on the application, the formulas used are of first order\nbsp{}cite:corke04_inert_visual_sensin_system_small_auton_helic,yeh05_model_contr_hydraul_actuat_two,yong16_high_speed_vertic_posit_stage, second order\nbsp{}cite:baerveldt97_low_cost_low_weigh_attit,stoten01_fusion_kinet_data_using_compos_filter,jensen13_basic_uas or even higher orders\nbsp{}cite:shaw90_bandw_enhan_posit_measur_using_measur_accel,zimmermann92_high_bandw_orien_measur_contr,stoten01_fusion_kinet_data_using_compos_filter,collette15_sensor_fusion_method_high_perfor,matichard15_seism_isolat_advan_ligo. \par
|
||||
|
||||
As the characteristics of the "super sensor" depends on the design of the complementary filters\nbsp{}cite:dehaeze19_compl_filter_shapin_using_synth, several optimization techniques were developed over the years.
|
||||
Some are based on the finding the optimal parameters of analytical formulas\nbsp{}cite:jensen13_basic_uas,min15_compl_filter_desig_angle_estim,becker15_compl_filter_desig_three_frequen_bands, while other are using convex optimization tools\nbsp{}cite:hua04_polyp_fir_compl_filter_contr_system,hua05_low_ligo such as linear matrix inequalities\nbsp{}cite:pascoal99_navig_system_desig_using_time.
|
||||
As the characteristics of the "super sensor" depends on the design of the complementary filters\nbsp{}cite:dehaeze19_compl_filter_shapin_using_synth, several optimization techniques have been developed over the years.
|
||||
Some are based on the finding optimal parameters of analytical formulas\nbsp{}cite:jensen13_basic_uas,min15_compl_filter_desig_angle_estim,becker15_compl_filter_desig_three_frequen_bands, while other are using convex optimization tools\nbsp{}cite:hua04_polyp_fir_compl_filter_contr_system,hua05_low_ligo such as linear matrix inequalities\nbsp{}cite:pascoal99_navig_system_desig_using_time.
|
||||
As shown in\nbsp{}cite:plummer06_optim_compl_filter_their_applic_motion_measur, the design of complementary filters can also be linked to the standard mixed-sensitivity control problem.
|
||||
Therefore, all the powerful tools developed for the classical control theory can also be used for the design of complementary filters.
|
||||
For instance, in\nbsp{}cite:jensen13_basic_uas the two gains of a Proportional Integral (PI) controller are optimized to minimize the noise of the super sensor. \par
|
||||
|
||||
** Problematic / gap in the research :ignore:
|
||||
|
||||
The common objective to all these complementary filters design methods is to obtain a super sensor that has desired characteristics, usually in terms of noise and dynamics.
|
||||
The common objective of all these complementary filters design methods is to obtain a super sensor that has desired characteristics, usually in terms of noise and dynamics.
|
||||
Moreover, as reported in\nbsp{}cite:zimmermann92_high_bandw_orien_measur_contr,plummer06_optim_compl_filter_their_applic_motion_measur, phase shifts and magnitude bumps of the "super sensors" dynamics can be observed if either the complementary filters are poorly designed or if the sensors are not well calibrated.
|
||||
Therefore, the robustness of the fusion is also of concerned when designing the complementary filters.
|
||||
Although many design methods of complementary filters have been proposed in the literature, no simple method that allows to specify the wanted super sensor characteristic while ensuring good fusion robustness has been proposed. \par
|
||||
Although many design methods of complementary filters have been proposed in the literature, no simple method that allows to specify the desired super sensor characteristic while ensuring good fusion robustness has been proposed. \par
|
||||
|
||||
** Describe the paper itself / the problem which is addressed :ignore:
|
||||
|
||||
@ -144,7 +153,7 @@ Based on that, this paper introduces a new way to design complementary filters u
|
||||
** Introduce Each part of the paper :ignore:
|
||||
|
||||
Section\nbsp{}ref:sec:requirements introduces the sensor fusion architecture and demonstrates how typical requirements can be linked to the complementary filters magnitudes.
|
||||
In section\nbsp{}ref:sec:hinf_method, the shaping of complementary filters is written as an $\mathcal{H}_\infty$ optimization problem using weighting functions, and the simplicity of the proposed method is illustrated with an example.
|
||||
In Section\nbsp{}ref:sec:hinf_method, the shaping of complementary filters is formulated as an $\mathcal{H}_\infty$ optimization problem using weighting functions, and the simplicity of the proposed method is illustrated with an example.
|
||||
The synthesis method is further validated in Section\nbsp{}ref:sec:application_ligo by designing complex complementary filters.
|
||||
Section\nbsp{}ref:sec:discussion compares the proposed synthesis method with the classical mixed-sensitivity synthesis, and extends it to the shaping of more than two complementary filters.
|
||||
|
||||
@ -167,7 +176,7 @@ The resulting sensor, termed as super sensor, can have larger bandwidth and bett
|
||||
This means that the super sensor provides an estimate $\hat{x}$ of $x$ which can be more accurate over a larger frequency band than the outputs of the individual sensors.
|
||||
|
||||
#+name: fig:sensor_fusion_overview
|
||||
#+caption: Schematic of a sensor fusion architecture using complementary filters
|
||||
#+caption: Schematic of a sensor fusion architecture using complementary filters.
|
||||
[[file:figs/sensor_fusion_overview.pdf]]
|
||||
|
||||
The complementary property of filters $H_1(s)$ and $H_2(s)$ implies that the sum of their transfer functions is equal to unity.
|
||||
@ -198,15 +207,15 @@ The sensor dynamics estimate $\hat{G}_1(s)$ can be a simple gain or a more compl
|
||||
\begin{subfigure}[b]{0.49\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_model.pdf}
|
||||
\caption{\label{fig:sensor_model} Basic sensor model consisting of a noise input $n_i$ and a linear time invariant transfer function $G_i(s)$}
|
||||
\caption{\label{fig:sensor_model} Basic sensor model consisting of a noise input $n_i$ and a linear time invariant transfer function $G_i(s)$.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\begin{subfigure}[b]{0.49\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_model_calibrated.pdf}
|
||||
\caption{\label{fig:sensor_model_calibrated} Normalized sensors using the inverse of an estimate $\hat{G}_i(s)$ of the sensor dynamics}
|
||||
\caption{\label{fig:sensor_model_calibrated} Normalized sensors using the inverse of an estimate $\hat{G}_i(s)$ of the sensor dynamics.}
|
||||
\end{subfigure}
|
||||
\caption{\label{fig:sensor_models}Sensor models with and without normalization}
|
||||
\caption{\label{fig:sensor_models}Sensor models with and without normalization.}
|
||||
\centering
|
||||
\end{figure}
|
||||
#+end_export
|
||||
@ -222,7 +231,7 @@ The super sensor output is therefore equal to:
|
||||
\end{equation}
|
||||
|
||||
#+name: fig:fusion_super_sensor
|
||||
#+caption: Sensor fusion architecture with two normalized sensors
|
||||
#+caption: Sensor fusion architecture with two normalized sensors.
|
||||
[[file:figs/fusion_super_sensor.pdf]]
|
||||
|
||||
** Noise Sensor Filtering
|
||||
@ -236,23 +245,23 @@ In this section, it is supposed that all the sensors are perfectly normalized, s
|
||||
|
||||
The effect of a non-perfect normalization will be discussed in the next section.
|
||||
|
||||
Provided eqref:eq:perfect_dynamics is verified, the super sensor output $\hat{x}$ is then:
|
||||
Provided\nbsp{}eqref:eq:perfect_dynamics is verified, the super sensor output $\hat{x}$ is then:
|
||||
#+name: eq:estimate_perfect_dyn
|
||||
\begin{equation}
|
||||
\hat{x} = x + H_1(s) n_1 + H_2(s) n_2
|
||||
\end{equation}
|
||||
|
||||
From eqref:eq:estimate_perfect_dyn, the complementary filters $H_1(s)$ and $H_2(s)$ are shown to only operate on the sensor's noises.
|
||||
From\nbsp{}eqref:eq:estimate_perfect_dyn, the complementary filters $H_1(s)$ and $H_2(s)$ are shown to only operate on the sensor's noises.
|
||||
Thus, this sensor fusion architecture permits to filter the noise of both sensors without introducing any distortion in the physical quantity to be measured.
|
||||
This is why the two filters must be complementary.
|
||||
|
||||
The estimation error $\delta x$, defined as the difference between the sensor output $\hat{x}$ and the measured quantity $x$, is computed for the super sensor eqref:eq:estimate_error.
|
||||
The estimation error $\delta x$, defined as the difference between the sensor output $\hat{x}$ and the measured quantity $x$, is computed for the super sensor\nbsp{}eqref:eq:estimate_error.
|
||||
#+name: eq:estimate_error
|
||||
\begin{equation}
|
||||
\delta x \triangleq \hat{x} - x = H_1(s) n_1 + H_2(s) n_2
|
||||
\end{equation}
|
||||
|
||||
As shown in eqref:eq:noise_filtering_psd, the Power Spectral Density (PSD) of the estimation error $\Phi_{\delta x}$ depends both on the norm of the two complementary filters and on the PSD of the noise sources $\Phi_{n_1}$ and $\Phi_{n_2}$.
|
||||
As shown in\nbsp{}eqref:eq:noise_filtering_psd, the Power Spectral Density (PSD) of the estimation error $\Phi_{\delta x}$ depends both on the norm of the two complementary filters and on the PSD of the noise sources $\Phi_{n_1}$ and $\Phi_{n_2}$.
|
||||
#+name: eq:noise_filtering_psd
|
||||
\begin{equation}
|
||||
\Phi_{\delta x}(\omega) = \left|H_1(j\omega)\right|^2 \Phi_{n_1}(\omega) + \left|H_2(j\omega)\right|^2 \Phi_{n_2}(\omega)
|
||||
@ -268,7 +277,7 @@ Therefore, by properly shaping the norm of the complementary filters, it is poss
|
||||
** Sensor Fusion Robustness
|
||||
<<sec:fusion_robustness>>
|
||||
|
||||
In practical systems the sensor normalization is not perfect and condition eqref:eq:perfect_dynamics is not verified.
|
||||
In practical systems the sensor normalization is not perfect and condition\nbsp{}eqref:eq:perfect_dynamics is not verified.
|
||||
|
||||
In order to study such imperfection, a multiplicative input uncertainty is added to the sensor dynamics (Fig.\nbsp{}ref:fig:sensor_model_uncertainty).
|
||||
The nominal model is taken as the estimated model for the normalization $\hat{G}_i(s)$, $\Delta_i$ is any stable transfer function satisfying $|\Delta_i(j\omega)| \le 1,\ \forall\omega$, and $w_i(s)$ is a weighting transfer function representing the magnitude of the uncertainty.
|
||||
@ -282,16 +291,16 @@ As the nominal sensor dynamics is taken as the normalized filter, the normalized
|
||||
\begin{subfigure}[b]{0.59\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_model_uncertainty.pdf}
|
||||
\caption{\label{fig:sensor_model_uncertainty} Sensor with multiplicative input uncertainty}
|
||||
\caption{\label{fig:sensor_model_uncertainty} Sensor with multiplicative input uncertainty.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\begin{subfigure}[b]{0.39\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_model_uncertainty_simplified.pdf}
|
||||
\caption{\label{fig:sensor_model_uncertainty_simplified} Simplified sensor model}
|
||||
\caption{\label{fig:sensor_model_uncertainty_simplified} Simplified sensor model.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\caption{\label{fig:sensor_models_uncertainty}Sensor models with dynamical uncertainty}
|
||||
\caption{\label{fig:sensor_models_uncertainty}Sensor models with dynamical uncertainty.}
|
||||
\centering
|
||||
\end{figure}
|
||||
#+end_export
|
||||
@ -299,10 +308,10 @@ As the nominal sensor dynamics is taken as the normalized filter, the normalized
|
||||
The sensor fusion architecture with two sensor models including dynamical uncertainty is shown in Fig.\nbsp{}ref:fig:sensor_fusion_dynamic_uncertainty.
|
||||
|
||||
#+name: fig:sensor_fusion_dynamic_uncertainty
|
||||
#+caption: Sensor fusion architecture with sensor dynamics uncertainty
|
||||
#+caption: Sensor fusion architecture with sensor dynamics uncertainty.
|
||||
[[file:figs/sensor_fusion_dynamic_uncertainty.pdf]]
|
||||
|
||||
The super sensor dynamics eqref:eq:super_sensor_dyn_uncertainty is no longer equal to $1$ and now depends on the sensor dynamical uncertainty weights $w_i(s)$ as well as on the complementary filters $H_i(s)$.
|
||||
The super sensor dynamics\nbsp{}eqref:eq:super_sensor_dyn_uncertainty is no longer equal to $1$ and now depends on the sensor dynamical uncertainty weights $w_i(s)$ as well as on the complementary filters $H_i(s)$.
|
||||
#+name: eq:super_sensor_dyn_uncertainty
|
||||
\begin{equation}
|
||||
\frac{\hat{x}}{x} = 1 + w_1(s) H_1(s) \Delta_1(s) + w_2(s) H_2(s) \Delta_2(s)
|
||||
@ -314,9 +323,9 @@ The dynamical uncertainty of the super sensor can be graphically represented in
|
||||
#+caption: Uncertainty region of the super sensor dynamics in the complex plane (grey circle). The contribution of both sensors 1 and 2 to the total uncertainty are represented respectively by a blue circle and a red circle. The frequency dependency $\omega$ is here omitted.
|
||||
[[file:figs/uncertainty_set_super_sensor.pdf]]
|
||||
|
||||
The super sensor dynamical uncertainty (i.e. the robustness of the fusion) clearly depends on the complementary filters norms.
|
||||
The super sensor dynamical uncertainty, and hence the robustness of the fusion, clearly depends on the complementary filters norms.
|
||||
For instance, the phase $\Delta\phi(\omega)$ added by the super sensor dynamics at frequency $\omega$ is bounded by $\Delta\phi_{\text{max}}(\omega)$ which can be found by drawing a tangent from the origin to the uncertainty circle of the super sensor (Fig.\nbsp{}ref:fig:uncertainty_set_super_sensor).
|
||||
Therefore, the phase uncertainty of the super sensor dynamics depends on the Complementary filters norms eqref:eq:max_phase_uncertainty.
|
||||
Therefore, the phase uncertainty of the super sensor dynamics depends on the Complementary filters norms\nbsp{}eqref:eq:max_phase_uncertainty.
|
||||
|
||||
#+name: eq:max_phase_uncertainty
|
||||
\begin{equation}
|
||||
@ -339,8 +348,8 @@ Finally, the synthesis method is validated on an simple example.
|
||||
** Synthesis Objective
|
||||
<<sec:synthesis_objective>>
|
||||
|
||||
The synthesis objective is to shape the norm of two filters $H_1(s)$ and $H_2(s)$ while ensuring their complementary property eqref:eq:comp_filter.
|
||||
This is equivalent as to finding proper and stable transfer functions $H_1(s)$ and $H_2(s)$ such that conditions eqref:eq:hinf_cond_complementarity, eqref:eq:hinf_cond_h1 and eqref:eq:hinf_cond_h2 are satisfied.
|
||||
The synthesis objective is to shape the norm of two filters $H_1(s)$ and $H_2(s)$ while ensuring their complementary property\nbsp{}eqref:eq:comp_filter.
|
||||
This is equivalent as to finding proper and stable transfer functions $H_1(s)$ and $H_2(s)$ such that conditions\nbsp{}eqref:eq:hinf_cond_complementarity, eqref:eq:hinf_cond_h1 and\nbsp{}eqref:eq:hinf_cond_h2 are satisfied.
|
||||
|
||||
#+name: eq:comp_filter_problem_form
|
||||
\begin{subequations}
|
||||
@ -357,7 +366,7 @@ $W_1(s)$ and $W_2(s)$ are two weighting transfer functions that are carefully ch
|
||||
|
||||
In this section, it is shown that the synthesis objective can be easily expressed as a standard $\mathcal{H}_\infty$ optimization problem and therefore solved using convenient tools readily available.
|
||||
|
||||
Consider the generalized plant $P(s)$ shown in Fig.\nbsp{}ref:fig:h_infinity_robust_fusion_plant and mathematically described by eqref:eq:generalized_plant.
|
||||
Consider the generalized plant $P(s)$ shown in Fig.\nbsp{}ref:fig:h_infinity_robust_fusion_plant and mathematically described by\nbsp{}eqref:eq:generalized_plant.
|
||||
|
||||
#+name: eq:generalized_plant
|
||||
\begin{equation}
|
||||
@ -369,28 +378,28 @@ Consider the generalized plant $P(s)$ shown in Fig.\nbsp{}ref:fig:h_infinity_rob
|
||||
\begin{subfigure}[t]{0.5\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/h_infinity_robust_fusion_plant.pdf}
|
||||
\caption{\label{fig:h_infinity_robust_fusion_plant} Generalized plant}
|
||||
\caption{\label{fig:h_infinity_robust_fusion_plant} Generalized plant.}
|
||||
\vfill
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\begin{subfigure}[t]{0.5\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/h_infinity_robust_fusion_fb.pdf}
|
||||
\caption{\label{fig:h_infinity_robust_fusion_fb} Generalized plant with the synthesized filter}
|
||||
\caption{\label{fig:h_infinity_robust_fusion_fb} Generalized plant with the synthesized filter.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\caption{\label{fig:h_infinity_robust_fusion} Architecture for the $\mathcal{H}_\infty$ synthesis of complementary filters}
|
||||
\caption{\label{fig:h_infinity_robust_fusion} Architecture for the $\mathcal{H}_\infty$ synthesis of complementary filters.}
|
||||
\centering
|
||||
\end{figure}
|
||||
#+end_export
|
||||
|
||||
Applying the standard $\mathcal{H}_\infty$ synthesis on the generalized plant $P(s)$ is then equivalent as finding a stable filter $H_2(s)$ which based on $v$, generates a signal $u$ such that the $\mathcal{H}_\infty$ norm of the system in Fig.\nbsp{}ref:fig:h_infinity_robust_fusion_fb from $w$ to $[z_1, \ z_2]$ is less than one eqref:eq:hinf_syn_obj.
|
||||
Applying the standard $\mathcal{H}_\infty$ synthesis on the generalized plant $P(s)$ is then equivalent as finding a stable filter $H_2(s)$ which based on $v$, generates a signal $u$ such that the $\mathcal{H}_\infty$ norm of the system in Fig.\nbsp{}ref:fig:h_infinity_robust_fusion_fb from $w$ to $[z_1, \ z_2]$ is less than one\nbsp{}eqref:eq:hinf_syn_obj.
|
||||
#+name: eq:hinf_syn_obj
|
||||
\begin{equation}
|
||||
\left\|\begin{matrix} \left(1 - H_2(s)\right) W_1(s) \\ H_2(s) W_2(s) \end{matrix}\right\|_\infty \le 1
|
||||
\end{equation}
|
||||
|
||||
By then defining $H_1(s)$ to be the complementary of $H_2(s)$ eqref:eq:definition_H1, the $\mathcal{H}_\infty$ synthesis objective becomes equivalent to eqref:eq:hinf_problem which ensure that eqref:eq:hinf_cond_h1 and eqref:eq:hinf_cond_h2 are satisfied.
|
||||
By then defining $H_1(s)$ to be the complementary of $H_2(s)$\nbsp{}eqref:eq:definition_H1, the $\mathcal{H}_\infty$ synthesis objective becomes equivalent to\nbsp{}eqref:eq:hinf_problem which ensure that\nbsp{}eqref:eq:hinf_cond_h1 and\nbsp{}eqref:eq:hinf_cond_h2 are satisfied.
|
||||
|
||||
#+name: eq:definition_H1
|
||||
\begin{equation}
|
||||
@ -402,7 +411,7 @@ By then defining $H_1(s)$ to be the complementary of $H_2(s)$ eqref:eq:definitio
|
||||
\left\|\begin{matrix} H_1(s) W_1(s) \\ H_2(s) W_2(s) \end{matrix}\right\|_\infty \le 1
|
||||
\end{equation}
|
||||
|
||||
Therefore, applying the $\mathcal{H}_\infty$ synthesis on the standard plant $P(s)$ eqref:eq:generalized_plant will generate two filters $H_2(s)$ and $H_1(s) \triangleq 1 - H_2(s)$ that are complementary eqref:eq:comp_filter_problem_form and such that there norms are bellow specified bounds [[eqref:eq:hinf_cond_h1]],eqref:eq:hinf_cond_h2.
|
||||
Therefore, applying the $\mathcal{H}_\infty$ synthesis on the standard plant $P(s)$\nbsp{}eqref:eq:generalized_plant will generate two filters $H_2(s)$ and $H_1(s) \triangleq 1 - H_2(s)$ that are complementary\nbsp{}eqref:eq:comp_filter_problem_form and such that there norms are bellow specified bounds [[eqref:eq:hinf_cond_h1]],\nbsp{}eqref:eq:hinf_cond_h2.
|
||||
|
||||
Note that there is not an equivalence between the $\mathcal{H}_\infty$ norm condition\nbsp{}eqref:eq:hinf_problem and the initial synthesis objectives\nbsp{}eqref:eq:hinf_cond_h1 and\nbsp{}eqref:eq:hinf_cond_h2, but only an implication.
|
||||
Hence, the optimization may be a little bit conservative with respect to the "set" of filters on which it is performed.
|
||||
@ -417,11 +426,11 @@ The proper design of these weighting functions is of primary importance for the
|
||||
|
||||
First, only proper and stable transfer functions should be used.
|
||||
Second, the order of the weighting functions should stay reasonably small in order to reduce the computational costs associated with the solving of the optimization problem and for the physical implementation of the filters (the order of the synthesized filters being equal to the sum of the weighting functions orders).
|
||||
Third, one should not forget the fundamental limitations imposed by the complementary property eqref:eq:comp_filter.
|
||||
Third, one should not forget the fundamental limitations imposed by the complementary property\nbsp{}eqref:eq:comp_filter.
|
||||
This implies for instance that $|H_1(j\omega)|$ and $|H_2(j\omega)|$ cannot be made small at the same frequency.
|
||||
|
||||
When designing complementary filters, it is usually desired to specify their slopes, their "blending" frequency and their maximum gains at low and high frequency.
|
||||
To help with the design of the weighting functions such that the above specification can be easily expressed, the formula eqref:eq:weight_formula is proposed.
|
||||
To help with the design of the weighting functions such that the above specification can be easily expressed, the formula\nbsp{}eqref:eq:weight_formula is proposed.
|
||||
|
||||
#+name: eq:weight_formula
|
||||
\begin{equation}
|
||||
@ -432,13 +441,13 @@ To help with the design of the weighting functions such that the above specifica
|
||||
}\right)^n
|
||||
\end{equation}
|
||||
|
||||
The parameters in formula eqref:eq:weight_formula are:
|
||||
- $G_0 = lim_{\omega \to 0} |W(j\omega)|$: the low frequency gain
|
||||
- $G_\infty = lim_{\omega \to \infty} |W(j\omega)|$: the high frequency gain
|
||||
The parameters in formula\nbsp{}eqref:eq:weight_formula are:
|
||||
- $G_0 = \lim_{\omega \to 0} |W(j\omega)|$: the low frequency gain
|
||||
- $G_\infty = \lim_{\omega \to \infty} |W(j\omega)|$: the high frequency gain
|
||||
- $G_c = |W(j\omega_c)|$: the gain at a specific frequency $\omega_c$ in $\si{rad/s}$.
|
||||
- $n$: the slope between high and low frequency. It also corresponds to the order of the weighting function.
|
||||
|
||||
The parameters $G_0$, $G_c$ and $G_\infty$ should either satisfy condition eqref:eq:cond_formula_1 or eqref:eq:cond_formula_2.
|
||||
The parameters $G_0$, $G_c$ and $G_\infty$ should either satisfy condition\nbsp{}eqref:eq:cond_formula_1 or\nbsp{}eqref:eq:cond_formula_2.
|
||||
#+name: eq:condition_params_formula
|
||||
\begin{subequations}
|
||||
\begin{align}
|
||||
@ -447,10 +456,10 @@ The parameters $G_0$, $G_c$ and $G_\infty$ should either satisfy condition eqref
|
||||
\end{align}
|
||||
\end{subequations}
|
||||
|
||||
An example of the obtained magnitude of a weighting function generated using eqref:eq:weight_formula is shown in Fig.\nbsp{}ref:fig:weight_formula.
|
||||
An example of the obtained magnitude of a weighting function generated using\nbsp{}eqref:eq:weight_formula is shown in Fig.\nbsp{}ref:fig:weight_formula.
|
||||
|
||||
#+name: fig:weight_formula
|
||||
#+caption: Magnitude of a weighting function generated using the proposed formula eqref:eq:weight_formula, $G_0 = 1e^{-3}$, $G_\infty = 10$, $\omega_c = \SI{10}{Hz}$, $G_c = 2$, $n = 3$
|
||||
#+caption: Magnitude of a weighting function generated using the proposed formula\nbsp{}eqref:eq:weight_formula, $G_0 = 1e^{-3}$, $G_\infty = 10$, $\omega_c = \SI{10}{Hz}$, $G_c = 2$, $n = 3$.
|
||||
[[file:figs/weight_formula.pdf]]
|
||||
|
||||
** Validation of the proposed synthesis method
|
||||
@ -462,7 +471,7 @@ The proposed methodology for the design of complementary filters is now applied
|
||||
- the slope of $|H_2(j\omega)|$ is $-3$ above $\SI{10}{Hz}$, its high frequency gain is $10^{-3}$
|
||||
|
||||
The first step is to translate the above requirements into the design of the weighting functions.
|
||||
The proposed formula eqref:eq:weight_formula is here used for such purpose.
|
||||
The proposed formula\nbsp{}eqref:eq:weight_formula is here used for such purpose.
|
||||
Parameters used are summarized in Table\nbsp{}ref:tab:weights_params.
|
||||
The inverse magnitudes of the designed weighting functions, which are representing the maximum allowed norms of the complementary filters, are shown by the dashed lines in Fig.\nbsp{}ref:fig:weights_W1_W2.
|
||||
|
||||
@ -481,20 +490,20 @@ Parameters & \(W_1(s)\) & \(W_2(s)\)\\
|
||||
\(n\) & \(2\) & \(3\)\\
|
||||
\bottomrule
|
||||
\end{tabularx}
|
||||
\captionof{table}{\label{tab:weights_params}Parameters used for weighting functions \(W_1(s)\) and \(W_2(s)\) using \eqref{eq:weight_formula}}
|
||||
\captionof{table}{\label{tab:weights_params}Parameters used for weighting functions \(W_1(s)\) and \(W_2(s)\) using~\eqref{eq:weight_formula}.}
|
||||
\end{minipage}
|
||||
\hfill
|
||||
\begin{minipage}[b]{0.49\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/weights_W1_W2.pdf}
|
||||
\caption{\label{fig:weights_W1_W2}Inverse magnitude of the weighting functions}
|
||||
\caption{\label{fig:weights_W1_W2}Inverse magnitude of the weighting functions.}
|
||||
\end{minipage}
|
||||
\end{figure}
|
||||
#+end_export
|
||||
|
||||
The $\mathcal{H}_\infty$ synthesis is then applied to the generalized plant of Fig.\nbsp{}ref:fig:h_infinity_robust_fusion_plant on efficiently solved in Matlab\nbsp{}cite:matlab20 using the Robust Control Toolbox.
|
||||
The filter $H_2(s)$ that minimizes the $\mathcal{H}_\infty$ norm between $w$ and $[z_1,\ z_2]^T$ is obtained.
|
||||
The $\mathcal{H}_\infty$ norm is here found to be close to one eqref:eq:hinf_synthesis_result which indicates that the synthesis is successful: the complementary filters norms are below the maximum specified upper bounds.
|
||||
The $\mathcal{H}_\infty$ norm is here found to be close to one\nbsp{}eqref:eq:hinf_synthesis_result which indicates that the synthesis is successful: the complementary filters norms are below the maximum specified upper bounds.
|
||||
This is confirmed by the bode plots of the obtained complementary filters in Fig.\nbsp{}ref:fig:hinf_filters_results.
|
||||
|
||||
#+name: eq:hinf_synthesis_result
|
||||
@ -502,7 +511,7 @@ This is confirmed by the bode plots of the obtained complementary filters in Fig
|
||||
\left\|\begin{matrix} \left(1 - H_2(s)\right) W_1(s) \\ H_2(s) W_2(s) \end{matrix}\right\|_\infty \approx 1
|
||||
\end{equation}
|
||||
|
||||
Their transfer functions in the Laplace domain are given in eqref:eq:hinf_synthesis_result_tf.
|
||||
Their transfer functions in the Laplace domain are given in\nbsp{}eqref:eq:hinf_synthesis_result_tf.
|
||||
As expected, the obtained filters are of order $5$, that is the sum of the weighting functions orders.
|
||||
|
||||
#+name: eq:hinf_synthesis_result_tf
|
||||
@ -515,7 +524,7 @@ As expected, the obtained filters are of order $5$, that is the sum of the weigh
|
||||
|
||||
|
||||
#+name: fig:hinf_filters_results
|
||||
#+caption: Bode plot of the obtained complementary filters
|
||||
#+caption: Bode plot of the obtained complementary filters.
|
||||
[[file:figs/hinf_filters_results.pdf]]
|
||||
|
||||
This simple example illustrates the fact that the proposed methodology for complementary filters shaping is easy to use and effective.
|
||||
@ -533,15 +542,15 @@ Then, another set of complementary filters ($L_1,H_1$) is used to merge the outp
|
||||
A simplified block diagram of the sensor fusion architecture is shown in Fig.\nbsp{}ref:fig:ligo_super_sensor_architecture.
|
||||
|
||||
#+name: fig:ligo_super_sensor_architecture
|
||||
#+caption: Simplified block diagram of the sensor blending strategy for the first stage at the LIGO\nbsp{}cite:hua04_low_ligo
|
||||
#+caption: Simplified block diagram of the sensor blending strategy for the first stage at the LIGO\nbsp{}cite:hua04_low_ligo.
|
||||
[[file:figs/ligo_super_sensor_architecture.pdf]]
|
||||
|
||||
The fusion of the position sensor at low frequency with the "inertial super sensor" at high frequency using the complementary filters ($L_1,H_1$) is done for several reasons, first of which is to give the super sensor a DC sensibility and therefore allow the feedback loop to have authority at zero frequency.
|
||||
The requirements on those filters are very tight and thus their design is complex and should be expressed as an optimization problem.
|
||||
The requirements on those filters are stringent and thus their design is complex and should be expressed as an optimization problem.
|
||||
|
||||
The approach used in\nbsp{}cite:hua05_low_ligo is to use FIR complementary filters and to write the synthesis as a convex optimization problem.
|
||||
After synthesis, the obtained FIR filters were found to be compliant with the requirements.
|
||||
However they are of very high order so their implementation is quite complex.
|
||||
However they are of high order so their implementation is quite complex.
|
||||
In this section, the effectiveness of the proposed complementary filter synthesis strategy is demonstrated on the same set of requirements.
|
||||
|
||||
** Complementary Filters Specifications
|
||||
@ -557,7 +566,7 @@ These specifications are therefore upper bounds on the complementary filters' ma
|
||||
They are physically represented in Fig.\nbsp{}ref:fig:fir_filter_ligo as well as the obtained magnitude of the FIR filters in\nbsp{}cite:hua05_low_ligo.
|
||||
|
||||
#+name: fig:fir_filter_ligo
|
||||
#+caption: Specifications and Bode plot of the obtained FIR complementary filters in\nbsp{}cite:hua05_low_ligo. The filters are here obtained using the SeDuMi Matlab toolbox\nbsp{}cite:sturm99_using_sedum
|
||||
#+caption: Specifications and Bode plot of the obtained FIR complementary filters in\nbsp{}cite:hua05_low_ligo. The filters are here obtained using the SeDuMi Matlab toolbox\nbsp{}cite:sturm99_using_sedum.
|
||||
[[file:figs/fir_filter_ligo.pdf]]
|
||||
|
||||
** Weighting Functions Design
|
||||
@ -571,7 +580,7 @@ For the one corresponding to the high pass filter $W_H(s)$, a $7^{\text{th}}$ or
|
||||
The magnitudes of the weighting functions are shown in Fig.\nbsp{}ref:fig:ligo_weights.
|
||||
|
||||
#+name: fig:ligo_weights
|
||||
#+caption: Specifications and weighting functions inverse magnitudes
|
||||
#+caption: Specifications and weighting functions inverse magnitudes.
|
||||
[[file:figs/ligo_weights.pdf]]
|
||||
|
||||
** $\mathcal{H}_\infty$ Synthesis of the complementary filters
|
||||
@ -580,11 +589,11 @@ The magnitudes of the weighting functions are shown in Fig.\nbsp{}ref:fig:ligo_w
|
||||
The proposed $\mathcal{H}_\infty$ synthesis is performed on the generalized plant shown in Fig.\nbsp{}ref:fig:h_infinity_robust_fusion_plant.
|
||||
After optimization, the $\mathcal{H}_\infty$ norm from $w$ to $[z_1,\ z_2]^T$ is found close to one indication successful synthesis.
|
||||
In Fig.\nbsp{}ref:fig:comp_fir_ligo_hinf, the bode plot of the obtained complementary filters are compared with the FIR filters of order 512 obtained in\nbsp{}cite:hua05_low_ligo.
|
||||
Even though the complementary filters using the $\mathcal{H}_\infty$ synthesis are of much lower order (order 27), they are found to be very close to the FIR filters.
|
||||
Even though the complementary filters using the $\mathcal{H}_\infty$ synthesis are of much lower order (order 27), they are found to be close to the FIR filters.
|
||||
This confirms the effectiveness of the proposed synthesis method even when the complementary filters are subject to complex requirements.
|
||||
|
||||
#+name: fig:comp_fir_ligo_hinf
|
||||
#+caption: Comparison of the FIR filters (dashed) designed in\nbsp{}cite:hua05_low_ligo with the filters obtained with $\mathcal{H}_\infty$ synthesis (solid)
|
||||
#+caption: Comparison of the FIR filters (dashed) designed in\nbsp{}cite:hua05_low_ligo with the filters obtained with $\mathcal{H}_\infty$ synthesis (solid).
|
||||
[[file:figs/comp_fir_ligo_hinf.pdf]]
|
||||
|
||||
* Discussion
|
||||
@ -594,14 +603,14 @@ This confirms the effectiveness of the proposed synthesis method even when the c
|
||||
** "Closed-Loop" complementary filters
|
||||
<<sec:closed_loop_complementary_filters>>
|
||||
An alternative way to implement complementary filters is by using a fundamental property of the classical feedback architecture shown in Fig.\nbsp{}ref:fig:feedback_sensor_fusion.
|
||||
This is for instance presented in cite:mahony05_compl_filter_desig_special_orthog,plummer06_optim_compl_filter_their_applic_motion_measur,jensen13_basic_uas.
|
||||
This is for instance presented in\nbsp{}cite:mahony05_compl_filter_desig_special_orthog,plummer06_optim_compl_filter_their_applic_motion_measur,jensen13_basic_uas.
|
||||
|
||||
#+name: fig:feedback_sensor_fusion
|
||||
#+caption: "Closed-Loop" complementary filters
|
||||
#+caption: "Closed-Loop" complementary filters.
|
||||
[[file:figs/feedback_sensor_fusion.pdf]]
|
||||
|
||||
Consider the feedback architecture of Fig.\nbsp{}ref:fig:feedback_sensor_fusion, with two inputs $\hat{x}_1$ and $\hat{x}_2$, and one output $\hat{x}$.
|
||||
The output $\hat{x}$ is linked to the inputs by eqref:eq:closed_loop_complementary_filters.
|
||||
The output $\hat{x}$ is linked to the inputs by\nbsp{}eqref:eq:closed_loop_complementary_filters.
|
||||
|
||||
#+name: eq:closed_loop_complementary_filters
|
||||
\begin{equation}
|
||||
@ -619,7 +628,7 @@ Therefore, provided that the closed-loop system is stable, the closed-loop syste
|
||||
Two sensors can then be merged as shown in Fig.\nbsp{}ref:fig:feedback_sensor_fusion_arch.
|
||||
|
||||
#+name: fig:feedback_sensor_fusion_arch
|
||||
#+caption: Classical feedback architecture used for sensor fusion
|
||||
#+caption: Classical feedback architecture used for sensor fusion.
|
||||
[[file:figs/feedback_sensor_fusion_arch.pdf]]
|
||||
|
||||
One of the main advantage of implementing and designing complementary filters using the feedback architecture of Fig.\nbsp{}ref:fig:feedback_sensor_fusion is that all the tools of the linear control theory can be applied for the design of the filters.
|
||||
@ -627,7 +636,7 @@ If one want to shape both $\frac{\hat{x}}{\hat{x}_1}(s) = S(s)$ and $\frac{\hat{
|
||||
|
||||
To do so, weighting functions $W_1(s)$ and $W_2(s)$ are added to respectively shape $S(s)$ and $T(s)$ (Fig.\nbsp{}ref:fig:feedback_synthesis_architecture).
|
||||
Then the system is re-organized to form the generalized plant $P_L(s)$ shown in Fig.\nbsp{}ref:fig:feedback_synthesis_architecture_generalized_plant.
|
||||
The $\mathcal{H}_\infty$ mixed-sensitivity synthesis can finally be performed by applying the $\mathcal{H}_\infty$ synthesis to the generalized plant $P_L(s)$ which is described by eqref:eq:generalized_plant_mixed_sensitivity.
|
||||
The $\mathcal{H}_\infty$ mixed-sensitivity synthesis can finally be performed by applying the $\mathcal{H}_\infty$ synthesis to the generalized plant $P_L(s)$ which is described by\nbsp{}eqref:eq:generalized_plant_mixed_sensitivity.
|
||||
|
||||
#+name: eq:generalized_plant_mixed_sensitivity
|
||||
\begin{equation}
|
||||
@ -637,7 +646,7 @@ The $\mathcal{H}_\infty$ mixed-sensitivity synthesis can finally be performed by
|
||||
\end{bmatrix}
|
||||
\end{equation}
|
||||
|
||||
The output of the synthesis is a filter $L(s)$ such that the "closed-loop" $\mathcal{H}_\infty$ norm from $[w_1,\ w_2]$ to $z$ is less than one eqref:eq:comp_filters_feedback_obj.
|
||||
The output of the synthesis is a filter $L(s)$ such that the "closed-loop" $\mathcal{H}_\infty$ norm from $[w_1,\ w_2]$ to $z$ is less than one\nbsp{}eqref:eq:comp_filters_feedback_obj.
|
||||
|
||||
#+name: eq:comp_filters_feedback_obj
|
||||
\begin{equation}
|
||||
@ -652,23 +661,23 @@ The sensor fusion can then be implemented as shown in Fig.\nbsp{}ref:fig:feedbac
|
||||
H_1(s) = \frac{1}{1 + L(s)}; \quad H_2(s) = \frac{L(s)}{1 + L(s)}
|
||||
\end{equation}
|
||||
|
||||
The two architecture are equivalent regarding their inputs/outputs relationships.
|
||||
The two architectures are equivalent regarding their inputs/outputs relationships.
|
||||
|
||||
#+begin_export latex
|
||||
\begin{figure}[htbp]
|
||||
\begin{subfigure}[t]{0.6\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/feedback_synthesis_architecture.pdf}
|
||||
\caption{\label{fig:feedback_synthesis_architecture} Feedback architecture with included weights}
|
||||
\caption{\label{fig:feedback_synthesis_architecture} Feedback architecture with included weights.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\begin{subfigure}[t]{0.4\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/feedback_synthesis_architecture_generalized_plant.pdf}
|
||||
\caption{\label{fig:feedback_synthesis_architecture_generalized_plant} Generalized plant}
|
||||
\caption{\label{fig:feedback_synthesis_architecture_generalized_plant} Generalized plant.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\caption{\label{fig:h_inf_mixed_sensitivity_synthesis} $\mathcal{H}_\infty$ mixed-sensitivity synthesis}
|
||||
\caption{\label{fig:h_inf_mixed_sensitivity_synthesis} $\mathcal{H}_\infty$ mixed-sensitivity synthesis.}
|
||||
\centering
|
||||
\end{figure}
|
||||
#+end_export
|
||||
@ -680,7 +689,7 @@ The "closed-loop" complementary filters are compared with the inverse magnitude
|
||||
The obtained "closed-loop" complementary filters are indeed equal to the ones obtained in Section\nbsp{}ref:sec:hinf_example.
|
||||
|
||||
#+name: fig:hinf_filters_results_mixed_sensitivity
|
||||
#+caption: Bode plot of the obtained complementary filters after $\mathcal{H}_\infty$ mixed-sensitivity synthesis
|
||||
#+caption: Bode plot of the obtained complementary filters after $\mathcal{H}_\infty$ mixed-sensitivity synthesis.
|
||||
[[file:figs/hinf_filters_results_mixed_sensitivity.pdf]]
|
||||
|
||||
** Synthesis of more than two Complementary Filters
|
||||
@ -701,15 +710,15 @@ A generalization of the proposed synthesis method of complementary filters is pr
|
||||
\begin{subfigure}[b]{0.59\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_fusion_three_sequential.pdf}
|
||||
\caption{\label{fig:sensor_fusion_three_sequential}Sequential fusion}
|
||||
\caption{\label{fig:sensor_fusion_three_sequential}Sequential fusion.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\begin{subfigure}[b]{0.39\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_fusion_three_parallel.pdf}
|
||||
\caption{\label{fig:sensor_fusion_three_parallel}Parallel fusion}
|
||||
\caption{\label{fig:sensor_fusion_three_parallel}Parallel fusion.}
|
||||
\end{subfigure}
|
||||
\caption{\label{fig:sensor_fusion_three}Possible sensor fusion architecture when more than two sensors are to be merged}
|
||||
\caption{\label{fig:sensor_fusion_three}Possible sensor fusion architecture when more than two sensors are to be merged.}
|
||||
\centering
|
||||
\end{figure}
|
||||
#+end_export
|
||||
@ -724,7 +733,7 @@ The synthesis objective is to compute a set of $n$ stable transfer functions $[H
|
||||
\end{subequations}
|
||||
where $[W_1(s),\ W_2(s),\ \dots,\ W_n(s)]$ are weighting transfer functions that are chosen to specify the maximum wanted norms of the complementary filters during the synthesis.
|
||||
|
||||
Such synthesis objective is very close to the one described in Section\nbsp{}ref:sec:synthesis_objective, and indeed the proposed synthesis method is a generalization of the one presented in Section\nbsp{}ref:sec:hinf_synthesis. \par
|
||||
Such synthesis objective is closely related to the one described in Section\nbsp{}ref:sec:synthesis_objective, and indeed the proposed synthesis method is a generalization of the one presented in Section\nbsp{}ref:sec:hinf_synthesis. \par
|
||||
|
||||
Before presenting the generalized synthesis method, the case with three sensors is presented.
|
||||
Consider the generalized plant $P_3(s)$ shown in Fig.\nbsp{}ref:fig:comp_filter_three_hinf_gen_plant which is also described by\nbsp{}eqref:eq:generalized_plant_three_filters.
|
||||
@ -739,17 +748,17 @@ Consider the generalized plant $P_3(s)$ shown in Fig.\nbsp{}ref:fig:comp_filter_
|
||||
\begin{subfigure}[t]{0.5\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/comp_filter_three_hinf_gen_plant.pdf}
|
||||
\caption{\label{fig:comp_filter_three_hinf_gen_plant} Generalized plant}
|
||||
\caption{\label{fig:comp_filter_three_hinf_gen_plant} Generalized plant.}
|
||||
\vfill
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\begin{subfigure}[t]{0.5\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/comp_filter_three_hinf_fb.pdf}
|
||||
\caption{\label{fig:comp_filter_three_hinf_fb} Generalized plant with the synthesized filter}
|
||||
\caption{\label{fig:comp_filter_three_hinf_fb} Generalized plant with the synthesized filter.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\caption{\label{fig:comp_filter_three_hinf} Architecture for the $\mathcal{H}_\infty$ synthesis of three complementary filters}
|
||||
\caption{\label{fig:comp_filter_three_hinf} Architecture for the $\mathcal{H}_\infty$ synthesis of three complementary filters.}
|
||||
\centering
|
||||
\end{figure}
|
||||
#+end_export
|
||||
@ -761,7 +770,7 @@ Applying the $\mathcal{H}_\infty$ synthesis on the generalized plant $P_3(s)$ is
|
||||
\left\| \begin{matrix} \left[1 - H_2(s) - H_3(s)\right] W_1(s) \\ H_2(s) W_2(s) \\ H_3(s) W_3(s) \end{matrix} \right\|_\infty \le 1
|
||||
\end{equation}
|
||||
|
||||
By defining $H_1(s) \triangleq 1 - H_2(s) - H_3(s)$, the proposed $\mathcal{H}_\infty$ synthesis solves the design problem eqref:eq:hinf_problem_gen with $n=3$. \par
|
||||
By defining $H_1(s) \triangleq 1 - H_2(s) - H_3(s)$, the proposed $\mathcal{H}_\infty$ synthesis solves the design problem\nbsp{}eqref:eq:hinf_problem_gen with $n=3$. \par
|
||||
|
||||
An example is given to validate the method where three sensors are used in different frequency bands.
|
||||
For instance a displacement sensor from DC up to $\SI{1}{Hz}$, a geophone from $1$ to $\SI{10}{Hz}$ and an accelerometer above $\SI{10}{Hz}$.
|
||||
@ -769,7 +778,7 @@ Three weighting functions are designed using formula\nbsp{}eqref:eq:weight_formu
|
||||
The $\mathcal{H}_\infty$ synthesis is performed on the generalized plant $P_3(s)$ and the bode plot of the obtained complementary filters are shown in Fig.\nbsp{}ref:fig:three_complementary_filters_results (solid lines). \par
|
||||
|
||||
#+name: fig:three_complementary_filters_results
|
||||
#+caption: Bode plot of the inverse weighting functions and of the three complementary filters obtained using the $\mathcal{H}_\infty$ synthesis
|
||||
#+caption: Bode plot of the inverse weighting functions and of the three complementary filters obtained using the $\mathcal{H}_\infty$ synthesis.
|
||||
[[file:figs/three_complementary_filters_results.pdf]]
|
||||
|
||||
Even though there might not be any practical application for a set of more than 3 complementary filters, it can still be designed using the same procedure.
|
||||
@ -791,15 +800,12 @@ A set of $n$ complementary filters can be shaped using the generalized plant $P_
|
||||
* Conclusion
|
||||
<<sec:conclusion>>
|
||||
|
||||
Sensors measuring a physical quantities are always subject to limitations both in terms of bandwidth or accuracy.
|
||||
Complementary filters can be used to fuse multiple sensors with different characteristics in order to combine their benefits and yield a better estimate of the measured physical quantity.
|
||||
|
||||
The sensor fusion robustness and the obtained super sensor noise can be linked to the magnitude of the complementary filters.
|
||||
In this paper, a synthesis method that enables the shaping of the complementary filters norms has been proposed.
|
||||
Several example were used to emphasize the simplicity and the effectiveness of the proposed method.
|
||||
Links with "closed-loop" complementary filters where highlighted, and the proposed method was generalized for the design of a set of more than two complementary filters.
|
||||
Several examples were used to emphasize the simplicity and the effectiveness of the proposed method.
|
||||
Links with "closed-loop" complementary filters were highlighted, and the proposed method was generalized for designing a set of more than two complementary filters.
|
||||
|
||||
Future work will aim at developing a complementary filter synthesis method that minimizes the super sensor noise while ensuring the robustness of the fusion.
|
||||
The future work will aim at developing a complementary filter synthesis method that minimizes the super sensor noise while ensuring the robustness of the fusion.
|
||||
|
||||
* Acknowledgment
|
||||
:PROPERTIES:
|
||||
@ -807,14 +813,13 @@ Future work will aim at developing a complementary filter synthesis method that
|
||||
:END:
|
||||
|
||||
This research benefited from a FRIA grant from the French Community of Belgium.
|
||||
This paper is based on a paper previously presented at the ICCMA conference\nbsp{}cite:dehaeze19_compl_filter_shapin_using_synth.
|
||||
|
||||
* Data Availability
|
||||
:PROPERTIES:
|
||||
:UNNUMBERED: t
|
||||
:END:
|
||||
|
||||
Matlab cite:matlab20 was used for this study.
|
||||
Matlab\nbsp{}cite:matlab20 was used for this study.
|
||||
The source code is available under a MIT License and archived in Zenodo\nbsp{}cite:dehaeze21_new_method_desig_compl_filter_code.
|
||||
|
||||
* Bibliography :ignore:
|
||||
|
||||
Binary file not shown.
@ -1,4 +1,4 @@
|
||||
% Created 2021-09-01 mer. 17:02
|
||||
% Created 2021-09-02 jeu. 10:02
|
||||
% Intended LaTeX compiler: pdflatex
|
||||
\documentclass[preprint, sort&compress]{elsarticle}
|
||||
\usepackage[utf8]{inputenc}
|
||||
@ -23,7 +23,7 @@
|
||||
\author[a2,a4]{Christophe Collette}
|
||||
\cortext[cor1]{Corresponding author. Email Address: thomas.dehaeze@esrf.fr}
|
||||
\address[a1]{European Synchrotron Radiation Facility, Grenoble, France}
|
||||
\address[a2]{University of Li\`{e}ge, Department of Aerospace and Mechanical Engineering, 4000 Li\`{e}ge, Belgium.}
|
||||
\address[a2]{University of Li\`{e}ge, PML, Department of Aerospace and Mechanical Engineering, 4000 Li\`{e}ge, Belgium.}
|
||||
\address[a3]{CSIR --- Structural Engineering Research Centre, Taramani, Chennai --- 600113, India.}
|
||||
\address[a4]{Universit\'{e} Libre de Bruxelles, Precision Mechatronics Laboratory, BEAMS Department, 1050 Brussels, Belgium.}
|
||||
\usepackage{tabularx}
|
||||
@ -34,7 +34,7 @@
|
||||
\hypersetup{colorlinks=true}
|
||||
\setlength{\parskip}{1em}
|
||||
\date{}
|
||||
\title{A new method of designing complementary filters for sensor fusion using the \(\mathcal{H}_\infty\) synthesis}
|
||||
\title{Designing complementary filters for sensor fusion using \(\mathcal{H}_\infty\) synthesis}
|
||||
\begin{document}
|
||||
|
||||
|
||||
@ -58,60 +58,57 @@ Sensor fusion \sep{} Complementary filters \sep{} \(\mathcal{H}_\infty\) synthes
|
||||
\end{frontmatter}
|
||||
|
||||
\section{Introduction}
|
||||
\label{sec:org2e9b248}
|
||||
\label{sec:introduction}
|
||||
Measuring a physical quantity using sensors is always subject to several limitations.
|
||||
First, the accuracy of the measurement will be affected by several noise sources, such as the electrical noise of the conditioning electronics being used.
|
||||
First, the accuracy of the measurement will be affected by several noise sources, such as electrical noise of the conditioning electronics being used.
|
||||
Second, the frequency range in which the measurement is relevant is bounded by the bandwidth of the sensor.
|
||||
One way to overcome these limitations is to combine several sensors using a technique called ``sensor fusion''~\cite{bendat57_optim_filter_indep_measur_two}.
|
||||
Fortunately, a wide variety of sensors exist, each with different characteristics.
|
||||
By carefully choosing the fused sensors, a so called ``super sensor'' is obtain that combines benefits of individual sensors and yields a better estimate of the measured physical quantity. \par
|
||||
By carefully choosing the fused sensors, a so called ``super sensor'' is obtained that combines benefits of individual sensors. \par
|
||||
In some situations, sensor fusion is used to increase the bandwidth of the measurement~\cite{shaw90_bandw_enhan_posit_measur_using_measur_accel,zimmermann92_high_bandw_orien_measur_contr,min15_compl_filter_desig_angle_estim}.
|
||||
For instance, in~\cite{shaw90_bandw_enhan_posit_measur_using_measur_accel}, the bandwidth of a position sensor is increased by fusing it with an accelerometer providing the high frequency motion information.
|
||||
For other applications, sensor fusion is used to obtain a estimate of the measured quantity with lower noise~\cite{hua05_low_ligo,hua04_polyp_fir_compl_filter_contr_system,plummer06_optim_compl_filter_their_applic_motion_measur,robert12_introd_random_signal_applied_kalman}.
|
||||
More recently, the fusion of sensors measuring different physical quantities has been proposed to obtained interesting properties for control~\cite{collette15_sensor_fusion_method_high_perfor,yong16_high_speed_vertic_posit_stage}.
|
||||
In~\cite{collette15_sensor_fusion_method_high_perfor}, an inertial sensor used for active vibration isolation is fused with a sensor collocated with the actuator to improve the stability margins of the feedback controller. \par
|
||||
For other applications, sensor fusion is used to obtain an estimate of the measured quantity with lower noise~\cite{hua05_low_ligo,hua04_polyp_fir_compl_filter_contr_system,plummer06_optim_compl_filter_their_applic_motion_measur,robert12_introd_random_signal_applied_kalman}.
|
||||
More recently, the fusion of sensors measuring different physical quantities has been proposed to obtain interesting properties for control~\cite{collette15_sensor_fusion_method_high_perfor,yong16_high_speed_vertic_posit_stage}.
|
||||
In~\cite{collette15_sensor_fusion_method_high_perfor}, an inertial sensor used for active vibration isolation is fused with a sensor collocated with the actuator for improving the stability margins of the feedback controller. \par
|
||||
Practical applications of sensor fusion are numerous.
|
||||
It is widely used for the attitude estimation of several autonomous vehicles such as unmanned aerial vehicle~\cite{baerveldt97_low_cost_low_weigh_attit,corke04_inert_visual_sensin_system_small_auton_helic,jensen13_basic_uas} and underwater vehicles~\cite{pascoal99_navig_system_desig_using_time,batista10_optim_posit_veloc_navig_filter_auton_vehic}.
|
||||
Naturally, it is of great benefits for high performance positioning control as shown in~\cite{shaw90_bandw_enhan_posit_measur_using_measur_accel,zimmermann92_high_bandw_orien_measur_contr,min15_compl_filter_desig_angle_estim,yong16_high_speed_vertic_posit_stage}.
|
||||
Sensor fusion was also shown to be a key technology to improve the performances of active vibration isolation systems~\cite{tjepkema12_sensor_fusion_activ_vibrat_isolat_precis_equip}.
|
||||
This is particularly apparent for the isolation stages of gravitational wave observer~\cite{collette15_sensor_fusion_method_high_perfor,heijningen18_low} such as the ones used at the LIGO~\cite{hua05_low_ligo,hua04_polyp_fir_compl_filter_contr_system} and at the VIRGO~\cite{lucia18_low_frequen_optim_perfor_advan}. \par
|
||||
Sensor fusion was also shown to be a key technology to improve the performance of active vibration isolation systems~\cite{tjepkema12_sensor_fusion_activ_vibrat_isolat_precis_equip}.
|
||||
Emblematic examples are the isolation stages of gravitational wave detectors~\cite{collette15_sensor_fusion_method_high_perfor,heijningen18_low} such as the ones used at the LIGO~\cite{hua05_low_ligo,hua04_polyp_fir_compl_filter_contr_system} and at the VIRGO~\cite{lucia18_low_frequen_optim_perfor_advan}. \par
|
||||
There are mainly two ways to perform sensor fusion: either using a set of complementary filters~\cite{anderson53_instr_approac_system_steer_comput} or using Kalman filtering~\cite{brown72_integ_navig_system_kalman_filter,odry18_kalman_filter_mobil_robot_attit_estim}.
|
||||
For sensor fusion applications, both methods are sharing many relationships~\cite{brown72_integ_navig_system_kalman_filter,higgins75_compar_compl_kalman_filter,robert12_introd_random_signal_applied_kalman,becker15_compl_filter_desig_three_frequen_bands}.
|
||||
However, for Kalman filtering, we are forced to make assumption about the probabilistic character of the sensor noises~\cite{robert12_introd_random_signal_applied_kalman} whereas it is not the case for complementary filters.
|
||||
Furthermore, the main advantages of complementary filters over Kalman filtering for sensor fusion are their very general applicability, their low computational cost~\cite{higgins75_compar_compl_kalman_filter}, and the fact that they are very intuitive as their effects can be easily interpreted in the frequency domain. \par
|
||||
However, for Kalman filtering, we are forced to make assumption about the probabilistic character of the sensor noises~\cite{robert12_introd_random_signal_applied_kalman} whereas it is not the case with complementary filters.
|
||||
Furthermore, the advantages of complementary filters over Kalman filtering for sensor fusion are their general applicability, their low computational cost~\cite{higgins75_compar_compl_kalman_filter}, and the fact that they are intuitive as their effects can be easily interpreted in the frequency domain. \par
|
||||
A set of filters is said to be complementary if the sum of their transfer functions is equal to one at all frequencies.
|
||||
For the earliest used of complementary filtering for sensor fusion, analog circuits were used to physically realize the filters~\cite{anderson53_instr_approac_system_steer_comput}.
|
||||
Analog complementary filters are still used today~\cite{yong16_high_speed_vertic_posit_stage,moore19_capac_instr_sensor_fusion_high_bandw_nanop}, but most of the time they are now implemented numerically as it allows for much more flexibility. \par
|
||||
In the early days of complementary filtering, analog circuits were employed to physically realize the filters~\cite{anderson53_instr_approac_system_steer_comput}.
|
||||
Analog complementary filters are still used today~\cite{yong16_high_speed_vertic_posit_stage,moore19_capac_instr_sensor_fusion_high_bandw_nanop}, but most of the time they are now implemented digitally as it allows for much more flexibility. \par
|
||||
|
||||
Several design methods have been developed over the years to optimize complementary filters.
|
||||
The easiest way to design complementary filters is to use analytical formulas.
|
||||
Depending on the application, the formulas used are of first order~\cite{corke04_inert_visual_sensin_system_small_auton_helic,yeh05_model_contr_hydraul_actuat_two,yong16_high_speed_vertic_posit_stage}, second order~\cite{baerveldt97_low_cost_low_weigh_attit,stoten01_fusion_kinet_data_using_compos_filter,jensen13_basic_uas} or even higher orders~\cite{shaw90_bandw_enhan_posit_measur_using_measur_accel,zimmermann92_high_bandw_orien_measur_contr,stoten01_fusion_kinet_data_using_compos_filter,collette15_sensor_fusion_method_high_perfor,matichard15_seism_isolat_advan_ligo}. \par
|
||||
|
||||
As the characteristics of the ``super sensor'' depends on the design of the complementary filters~\cite{dehaeze19_compl_filter_shapin_using_synth}, several optimization techniques were developed over the years.
|
||||
Some are based on the finding the optimal parameters of analytical formulas~\cite{jensen13_basic_uas,min15_compl_filter_desig_angle_estim,becker15_compl_filter_desig_three_frequen_bands}, while other are using convex optimization tools~\cite{hua04_polyp_fir_compl_filter_contr_system,hua05_low_ligo} such as linear matrix inequalities~\cite{pascoal99_navig_system_desig_using_time}.
|
||||
As the characteristics of the ``super sensor'' depends on the design of the complementary filters~\cite{dehaeze19_compl_filter_shapin_using_synth}, several optimization techniques have been developed over the years.
|
||||
Some are based on the finding optimal parameters of analytical formulas~\cite{jensen13_basic_uas,min15_compl_filter_desig_angle_estim,becker15_compl_filter_desig_three_frequen_bands}, while other are using convex optimization tools~\cite{hua04_polyp_fir_compl_filter_contr_system,hua05_low_ligo} such as linear matrix inequalities~\cite{pascoal99_navig_system_desig_using_time}.
|
||||
As shown in~\cite{plummer06_optim_compl_filter_their_applic_motion_measur}, the design of complementary filters can also be linked to the standard mixed-sensitivity control problem.
|
||||
Therefore, all the powerful tools developed for the classical control theory can also be used for the design of complementary filters.
|
||||
For instance, in~\cite{jensen13_basic_uas} the two gains of a Proportional Integral (PI) controller are optimized to minimize the noise of the super sensor. \par
|
||||
The common objective to all these complementary filters design methods is to obtain a super sensor that has desired characteristics, usually in terms of noise and dynamics.
|
||||
The common objective of all these complementary filters design methods is to obtain a super sensor that has desired characteristics, usually in terms of noise and dynamics.
|
||||
Moreover, as reported in~\cite{zimmermann92_high_bandw_orien_measur_contr,plummer06_optim_compl_filter_their_applic_motion_measur}, phase shifts and magnitude bumps of the ``super sensors'' dynamics can be observed if either the complementary filters are poorly designed or if the sensors are not well calibrated.
|
||||
Therefore, the robustness of the fusion is also of concerned when designing the complementary filters.
|
||||
Although many design methods of complementary filters have been proposed in the literature, no simple method that allows to specify the wanted super sensor characteristic while ensuring good fusion robustness has been proposed. \par
|
||||
Although many design methods of complementary filters have been proposed in the literature, no simple method that allows to specify the desired super sensor characteristic while ensuring good fusion robustness has been proposed. \par
|
||||
Fortunately, both the robustness of the fusion and the super sensor characteristics can be linked to the magnitude of the complementary filters~\cite{dehaeze19_compl_filter_shapin_using_synth}.
|
||||
Based on that, this paper introduces a new way to design complementary filters using the \(\mathcal{H}_\infty\) synthesis which allows to shape the magnitude of the complementary filters in an easy and intuitive way. \par
|
||||
Section~\ref{sec:requirements} introduces the sensor fusion architecture and demonstrates how typical requirements can be linked to the complementary filters magnitudes.
|
||||
In section~\ref{sec:hinf_method}, the shaping of complementary filters is written as an \(\mathcal{H}_\infty\) optimization problem using weighting functions, and the simplicity of the proposed method is illustrated with an example.
|
||||
In Section~\ref{sec:hinf_method}, the shaping of complementary filters is formulated as an \(\mathcal{H}_\infty\) optimization problem using weighting functions, and the simplicity of the proposed method is illustrated with an example.
|
||||
The synthesis method is further validated in Section~\ref{sec:application_ligo} by designing complex complementary filters.
|
||||
Section~\ref{sec:discussion} compares the proposed synthesis method with the classical mixed-sensitivity synthesis, and extends it to the shaping of more than two complementary filters.
|
||||
|
||||
\section{Sensor Fusion and Complementary Filters Requirements}
|
||||
\label{sec:org11c2c5d}
|
||||
\label{sec:requirements}
|
||||
Complementary filters provides a framework for fusing signals from different sensors.
|
||||
As the effectiveness of the fusion depends on the proper design of the complementary filters, they are expected to fulfill certain requirements.
|
||||
These requirements are discussed in this section.
|
||||
\subsection{Sensor Fusion Architecture}
|
||||
\label{sec:orgf188ee8}
|
||||
\label{sec:sensor_fusion}
|
||||
|
||||
A general sensor fusion architecture using complementary filters is shown in Fig.~\ref{fig:sensor_fusion_overview} where several sensors (here two) are measuring the same physical quantity \(x\).
|
||||
@ -124,7 +121,7 @@ This means that the super sensor provides an estimate \(\hat{x}\) of \(x\) which
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_fusion_overview.pdf}
|
||||
\caption{\label{fig:sensor_fusion_overview}Schematic of a sensor fusion architecture using complementary filters}
|
||||
\caption{\label{fig:sensor_fusion_overview}Schematic of a sensor fusion architecture using complementary filters.}
|
||||
\end{figure}
|
||||
|
||||
The complementary property of filters \(H_1(s)\) and \(H_2(s)\) implies that the sum of their transfer functions is equal to unity.
|
||||
@ -138,7 +135,6 @@ Therefore, a pair of complementary filter needs to satisfy the following conditi
|
||||
It will soon become clear why the complementary property is important for the sensor fusion architecture.
|
||||
|
||||
\subsection{Sensor Models and Sensor Normalization}
|
||||
\label{sec:orgb672f70}
|
||||
\label{sec:sensor_models}
|
||||
|
||||
In order to study such sensor fusion architecture, a model for the sensors is required.
|
||||
@ -155,15 +151,15 @@ The sensor dynamics estimate \(\hat{G}_1(s)\) can be a simple gain or a more com
|
||||
\begin{subfigure}[b]{0.49\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_model.pdf}
|
||||
\caption{\label{fig:sensor_model} Basic sensor model consisting of a noise input $n_i$ and a linear time invariant transfer function $G_i(s)$}
|
||||
\caption{\label{fig:sensor_model} Basic sensor model consisting of a noise input $n_i$ and a linear time invariant transfer function $G_i(s)$.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\begin{subfigure}[b]{0.49\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_model_calibrated.pdf}
|
||||
\caption{\label{fig:sensor_model_calibrated} Normalized sensors using the inverse of an estimate $\hat{G}_i(s)$ of the sensor dynamics}
|
||||
\caption{\label{fig:sensor_model_calibrated} Normalized sensors using the inverse of an estimate $\hat{G}_i(s)$ of the sensor dynamics.}
|
||||
\end{subfigure}
|
||||
\caption{\label{fig:sensor_models}Sensor models with and without normalization}
|
||||
\caption{\label{fig:sensor_models}Sensor models with and without normalization.}
|
||||
\centering
|
||||
\end{figure}
|
||||
|
||||
@ -180,11 +176,10 @@ The super sensor output is therefore equal to:
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/fusion_super_sensor.pdf}
|
||||
\caption{\label{fig:fusion_super_sensor}Sensor fusion architecture with two normalized sensors}
|
||||
\caption{\label{fig:fusion_super_sensor}Sensor fusion architecture with two normalized sensors.}
|
||||
\end{figure}
|
||||
|
||||
\subsection{Noise Sensor Filtering}
|
||||
\label{sec:org6621cf0}
|
||||
\label{sec:noise_filtering}
|
||||
|
||||
In this section, it is supposed that all the sensors are perfectly normalized, such that:
|
||||
@ -195,23 +190,23 @@ In this section, it is supposed that all the sensors are perfectly normalized, s
|
||||
|
||||
The effect of a non-perfect normalization will be discussed in the next section.
|
||||
|
||||
Provided \eqref{eq:perfect_dynamics} is verified, the super sensor output \(\hat{x}\) is then:
|
||||
Provided~\eqref{eq:perfect_dynamics} is verified, the super sensor output \(\hat{x}\) is then:
|
||||
\begin{equation}
|
||||
\label{eq:estimate_perfect_dyn}
|
||||
\hat{x} = x + H_1(s) n_1 + H_2(s) n_2
|
||||
\end{equation}
|
||||
|
||||
From \eqref{eq:estimate_perfect_dyn}, the complementary filters \(H_1(s)\) and \(H_2(s)\) are shown to only operate on the sensor's noises.
|
||||
From~\eqref{eq:estimate_perfect_dyn}, the complementary filters \(H_1(s)\) and \(H_2(s)\) are shown to only operate on the sensor's noises.
|
||||
Thus, this sensor fusion architecture permits to filter the noise of both sensors without introducing any distortion in the physical quantity to be measured.
|
||||
This is why the two filters must be complementary.
|
||||
|
||||
The estimation error \(\delta x\), defined as the difference between the sensor output \(\hat{x}\) and the measured quantity \(x\), is computed for the super sensor \eqref{eq:estimate_error}.
|
||||
The estimation error \(\delta x\), defined as the difference between the sensor output \(\hat{x}\) and the measured quantity \(x\), is computed for the super sensor~\eqref{eq:estimate_error}.
|
||||
\begin{equation}
|
||||
\label{eq:estimate_error}
|
||||
\delta x \triangleq \hat{x} - x = H_1(s) n_1 + H_2(s) n_2
|
||||
\end{equation}
|
||||
|
||||
As shown in \eqref{eq:noise_filtering_psd}, the Power Spectral Density (PSD) of the estimation error \(\Phi_{\delta x}\) depends both on the norm of the two complementary filters and on the PSD of the noise sources \(\Phi_{n_1}\) and \(\Phi_{n_2}\).
|
||||
As shown in~\eqref{eq:noise_filtering_psd}, the Power Spectral Density (PSD) of the estimation error \(\Phi_{\delta x}\) depends both on the norm of the two complementary filters and on the PSD of the noise sources \(\Phi_{n_1}\) and \(\Phi_{n_2}\).
|
||||
\begin{equation}
|
||||
\label{eq:noise_filtering_psd}
|
||||
\Phi_{\delta x}(\omega) = \left|H_1(j\omega)\right|^2 \Phi_{n_1}(\omega) + \left|H_2(j\omega)\right|^2 \Phi_{n_2}(\omega)
|
||||
@ -225,10 +220,9 @@ In such case, to lower the noise of the super sensor, the norm \(|H_1(j\omega)|\
|
||||
Therefore, by properly shaping the norm of the complementary filters, it is possible to minimize the noise of the super sensor noise.
|
||||
|
||||
\subsection{Sensor Fusion Robustness}
|
||||
\label{sec:orge7e30ce}
|
||||
\label{sec:fusion_robustness}
|
||||
|
||||
In practical systems the sensor normalization is not perfect and condition \eqref{eq:perfect_dynamics} is not verified.
|
||||
In practical systems the sensor normalization is not perfect and condition~\eqref{eq:perfect_dynamics} is not verified.
|
||||
|
||||
In order to study such imperfection, a multiplicative input uncertainty is added to the sensor dynamics (Fig.~\ref{fig:sensor_model_uncertainty}).
|
||||
The nominal model is taken as the estimated model for the normalization \(\hat{G}_i(s)\), \(\Delta_i\) is any stable transfer function satisfying \(|\Delta_i(j\omega)| \le 1,\ \forall\omega\), and \(w_i(s)\) is a weighting transfer function representing the magnitude of the uncertainty.
|
||||
@ -241,16 +235,16 @@ As the nominal sensor dynamics is taken as the normalized filter, the normalized
|
||||
\begin{subfigure}[b]{0.59\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_model_uncertainty.pdf}
|
||||
\caption{\label{fig:sensor_model_uncertainty} Sensor with multiplicative input uncertainty}
|
||||
\caption{\label{fig:sensor_model_uncertainty} Sensor with multiplicative input uncertainty.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\begin{subfigure}[b]{0.39\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_model_uncertainty_simplified.pdf}
|
||||
\caption{\label{fig:sensor_model_uncertainty_simplified} Simplified sensor model}
|
||||
\caption{\label{fig:sensor_model_uncertainty_simplified} Simplified sensor model.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\caption{\label{fig:sensor_models_uncertainty}Sensor models with dynamical uncertainty}
|
||||
\caption{\label{fig:sensor_models_uncertainty}Sensor models with dynamical uncertainty.}
|
||||
\centering
|
||||
\end{figure}
|
||||
|
||||
@ -259,10 +253,10 @@ The sensor fusion architecture with two sensor models including dynamical uncert
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_fusion_dynamic_uncertainty.pdf}
|
||||
\caption{\label{fig:sensor_fusion_dynamic_uncertainty}Sensor fusion architecture with sensor dynamics uncertainty}
|
||||
\caption{\label{fig:sensor_fusion_dynamic_uncertainty}Sensor fusion architecture with sensor dynamics uncertainty.}
|
||||
\end{figure}
|
||||
|
||||
The super sensor dynamics \eqref{eq:super_sensor_dyn_uncertainty} is no longer equal to \(1\) and now depends on the sensor dynamical uncertainty weights \(w_i(s)\) as well as on the complementary filters \(H_i(s)\).
|
||||
The super sensor dynamics~\eqref{eq:super_sensor_dyn_uncertainty} is no longer equal to \(1\) and now depends on the sensor dynamical uncertainty weights \(w_i(s)\) as well as on the complementary filters \(H_i(s)\).
|
||||
\begin{equation}
|
||||
\label{eq:super_sensor_dyn_uncertainty}
|
||||
\frac{\hat{x}}{x} = 1 + w_1(s) H_1(s) \Delta_1(s) + w_2(s) H_2(s) \Delta_2(s)
|
||||
@ -276,9 +270,9 @@ The dynamical uncertainty of the super sensor can be graphically represented in
|
||||
\caption{\label{fig:uncertainty_set_super_sensor}Uncertainty region of the super sensor dynamics in the complex plane (grey circle). The contribution of both sensors 1 and 2 to the total uncertainty are represented respectively by a blue circle and a red circle. The frequency dependency \(\omega\) is here omitted.}
|
||||
\end{figure}
|
||||
|
||||
The super sensor dynamical uncertainty (i.e. the robustness of the fusion) clearly depends on the complementary filters norms.
|
||||
The super sensor dynamical uncertainty, and hence the robustness of the fusion, clearly depends on the complementary filters norms.
|
||||
For instance, the phase \(\Delta\phi(\omega)\) added by the super sensor dynamics at frequency \(\omega\) is bounded by \(\Delta\phi_{\text{max}}(\omega)\) which can be found by drawing a tangent from the origin to the uncertainty circle of the super sensor (Fig.~\ref{fig:uncertainty_set_super_sensor}).
|
||||
Therefore, the phase uncertainty of the super sensor dynamics depends on the Complementary filters norms \eqref{eq:max_phase_uncertainty}.
|
||||
Therefore, the phase uncertainty of the super sensor dynamics depends on the Complementary filters norms~\eqref{eq:max_phase_uncertainty}.
|
||||
|
||||
\begin{equation}
|
||||
\label{eq:max_phase_uncertainty}
|
||||
@ -289,7 +283,6 @@ As it is generally desired to limit the maximum phase added by the super sensor,
|
||||
Typically, the norm of the complementary filter \(|H_i(j\omega)|\) should be made small when \(|w_i(j\omega)|\) is large, i.e., at frequencies where the sensor dynamics is uncertain.
|
||||
|
||||
\section{Complementary Filters Shaping}
|
||||
\label{sec:orgffe483d}
|
||||
\label{sec:hinf_method}
|
||||
As shown in Section~\ref{sec:requirements}, the noise and robustness of the super sensor are a function of the complementary filters norms.
|
||||
Therefore, a complementary filters synthesis method that allows to shape their norms would be of great use.
|
||||
@ -297,11 +290,10 @@ In this section, such synthesis is proposed by writing the synthesis objective a
|
||||
As weighting functions are used to represent the wanted complementary filters shapes during the synthesis, the proper design of weighting functions is discussed.
|
||||
Finally, the synthesis method is validated on an simple example.
|
||||
\subsection{Synthesis Objective}
|
||||
\label{sec:orgc020e5d}
|
||||
\label{sec:synthesis_objective}
|
||||
|
||||
The synthesis objective is to shape the norm of two filters \(H_1(s)\) and \(H_2(s)\) while ensuring their complementary property \eqref{eq:comp_filter}.
|
||||
This is equivalent as to finding proper and stable transfer functions \(H_1(s)\) and \(H_2(s)\) such that conditions \eqref{eq:hinf_cond_complementarity}, \eqref{eq:hinf_cond_h1} and \eqref{eq:hinf_cond_h2} are satisfied.
|
||||
The synthesis objective is to shape the norm of two filters \(H_1(s)\) and \(H_2(s)\) while ensuring their complementary property~\eqref{eq:comp_filter}.
|
||||
This is equivalent as to finding proper and stable transfer functions \(H_1(s)\) and \(H_2(s)\) such that conditions~\eqref{eq:hinf_cond_complementarity}, \eqref{eq:hinf_cond_h1} and~\eqref{eq:hinf_cond_h2} are satisfied.
|
||||
|
||||
\begin{subequations}
|
||||
\label{eq:comp_filter_problem_form}
|
||||
@ -314,12 +306,11 @@ This is equivalent as to finding proper and stable transfer functions \(H_1(s)\)
|
||||
\(W_1(s)\) and \(W_2(s)\) are two weighting transfer functions that are carefully chosen to specify the maximum wanted norms of the complementary filters during the synthesis.
|
||||
|
||||
\subsection{Shaping of Complementary Filters using \(\mathcal{H}_\infty\) synthesis}
|
||||
\label{sec:org6c86cb9}
|
||||
\label{sec:hinf_synthesis}
|
||||
|
||||
In this section, it is shown that the synthesis objective can be easily expressed as a standard \(\mathcal{H}_\infty\) optimization problem and therefore solved using convenient tools readily available.
|
||||
|
||||
Consider the generalized plant \(P(s)\) shown in Fig.~\ref{fig:h_infinity_robust_fusion_plant} and mathematically described by \eqref{eq:generalized_plant}.
|
||||
Consider the generalized plant \(P(s)\) shown in Fig.~\ref{fig:h_infinity_robust_fusion_plant} and mathematically described by~\eqref{eq:generalized_plant}.
|
||||
|
||||
\begin{equation}
|
||||
\label{eq:generalized_plant}
|
||||
@ -330,27 +321,27 @@ Consider the generalized plant \(P(s)\) shown in Fig.~\ref{fig:h_infinity_robust
|
||||
\begin{subfigure}[t]{0.5\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/h_infinity_robust_fusion_plant.pdf}
|
||||
\caption{\label{fig:h_infinity_robust_fusion_plant} Generalized plant}
|
||||
\caption{\label{fig:h_infinity_robust_fusion_plant} Generalized plant.}
|
||||
\vfill
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\begin{subfigure}[t]{0.5\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/h_infinity_robust_fusion_fb.pdf}
|
||||
\caption{\label{fig:h_infinity_robust_fusion_fb} Generalized plant with the synthesized filter}
|
||||
\caption{\label{fig:h_infinity_robust_fusion_fb} Generalized plant with the synthesized filter.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\caption{\label{fig:h_infinity_robust_fusion} Architecture for the $\mathcal{H}_\infty$ synthesis of complementary filters}
|
||||
\caption{\label{fig:h_infinity_robust_fusion} Architecture for the $\mathcal{H}_\infty$ synthesis of complementary filters.}
|
||||
\centering
|
||||
\end{figure}
|
||||
|
||||
Applying the standard \(\mathcal{H}_\infty\) synthesis on the generalized plant \(P(s)\) is then equivalent as finding a stable filter \(H_2(s)\) which based on \(v\), generates a signal \(u\) such that the \(\mathcal{H}_\infty\) norm of the system in Fig.~\ref{fig:h_infinity_robust_fusion_fb} from \(w\) to \([z_1, \ z_2]\) is less than one \eqref{eq:hinf_syn_obj}.
|
||||
Applying the standard \(\mathcal{H}_\infty\) synthesis on the generalized plant \(P(s)\) is then equivalent as finding a stable filter \(H_2(s)\) which based on \(v\), generates a signal \(u\) such that the \(\mathcal{H}_\infty\) norm of the system in Fig.~\ref{fig:h_infinity_robust_fusion_fb} from \(w\) to \([z_1, \ z_2]\) is less than one~\eqref{eq:hinf_syn_obj}.
|
||||
\begin{equation}
|
||||
\label{eq:hinf_syn_obj}
|
||||
\left\|\begin{matrix} \left(1 - H_2(s)\right) W_1(s) \\ H_2(s) W_2(s) \end{matrix}\right\|_\infty \le 1
|
||||
\end{equation}
|
||||
|
||||
By then defining \(H_1(s)\) to be the complementary of \(H_2(s)\) \eqref{eq:definition_H1}, the \(\mathcal{H}_\infty\) synthesis objective becomes equivalent to \eqref{eq:hinf_problem} which ensure that \eqref{eq:hinf_cond_h1} and \eqref{eq:hinf_cond_h2} are satisfied.
|
||||
By then defining \(H_1(s)\) to be the complementary of \$H\_2(s)\$~\eqref{eq:definition_H1}, the \(\mathcal{H}_\infty\) synthesis objective becomes equivalent to~\eqref{eq:hinf_problem} which ensure that~\eqref{eq:hinf_cond_h1} and~\eqref{eq:hinf_cond_h2} are satisfied.
|
||||
|
||||
\begin{equation}
|
||||
\label{eq:definition_H1}
|
||||
@ -362,7 +353,7 @@ By then defining \(H_1(s)\) to be the complementary of \(H_2(s)\) \eqref{eq:defi
|
||||
\left\|\begin{matrix} H_1(s) W_1(s) \\ H_2(s) W_2(s) \end{matrix}\right\|_\infty \le 1
|
||||
\end{equation}
|
||||
|
||||
Therefore, applying the \(\mathcal{H}_\infty\) synthesis on the standard plant \(P(s)\) \eqref{eq:generalized_plant} will generate two filters \(H_2(s)\) and \(H_1(s) \triangleq 1 - H_2(s)\) that are complementary \eqref{eq:comp_filter_problem_form} and such that there norms are bellow specified bounds \eqref{eq:hinf_cond_h1},\eqref{eq:hinf_cond_h2}.
|
||||
Therefore, applying the \(\mathcal{H}_\infty\) synthesis on the standard plant \$P(s)\$~\eqref{eq:generalized_plant} will generate two filters \(H_2(s)\) and \(H_1(s) \triangleq 1 - H_2(s)\) that are complementary~\eqref{eq:comp_filter_problem_form} and such that there norms are bellow specified bounds \eqref{eq:hinf_cond_h1},~\eqref{eq:hinf_cond_h2}.
|
||||
|
||||
Note that there is not an equivalence between the \(\mathcal{H}_\infty\) norm condition~\eqref{eq:hinf_problem} and the initial synthesis objectives~\eqref{eq:hinf_cond_h1} and~\eqref{eq:hinf_cond_h2}, but only an implication.
|
||||
Hence, the optimization may be a little bit conservative with respect to the ``set'' of filters on which it is performed.
|
||||
@ -370,7 +361,6 @@ There might be solutions were the objectives~\eqref{eq:hinf_cond_h1} and~\eqref{
|
||||
In practice, this is however not an found to be an issue.
|
||||
|
||||
\subsection{Weighting Functions Design}
|
||||
\label{sec:orgf11126f}
|
||||
\label{sec:hinf_weighting_func}
|
||||
|
||||
Weighting functions are used during the synthesis to specify the maximum allowed norms of the complementary filters.
|
||||
@ -378,11 +368,11 @@ The proper design of these weighting functions is of primary importance for the
|
||||
|
||||
First, only proper and stable transfer functions should be used.
|
||||
Second, the order of the weighting functions should stay reasonably small in order to reduce the computational costs associated with the solving of the optimization problem and for the physical implementation of the filters (the order of the synthesized filters being equal to the sum of the weighting functions orders).
|
||||
Third, one should not forget the fundamental limitations imposed by the complementary property \eqref{eq:comp_filter}.
|
||||
Third, one should not forget the fundamental limitations imposed by the complementary property~\eqref{eq:comp_filter}.
|
||||
This implies for instance that \(|H_1(j\omega)|\) and \(|H_2(j\omega)|\) cannot be made small at the same frequency.
|
||||
|
||||
When designing complementary filters, it is usually desired to specify their slopes, their ``blending'' frequency and their maximum gains at low and high frequency.
|
||||
To help with the design of the weighting functions such that the above specification can be easily expressed, the formula \eqref{eq:weight_formula} is proposed.
|
||||
To help with the design of the weighting functions such that the above specification can be easily expressed, the formula~\eqref{eq:weight_formula} is proposed.
|
||||
|
||||
\begin{equation}
|
||||
\label{eq:weight_formula}
|
||||
@ -393,15 +383,15 @@ To help with the design of the weighting functions such that the above specifica
|
||||
}\right)^n
|
||||
\end{equation}
|
||||
|
||||
The parameters in formula \eqref{eq:weight_formula} are:
|
||||
The parameters in formula~\eqref{eq:weight_formula} are:
|
||||
\begin{itemize}
|
||||
\item \(G_0 = lim_{\omega \to 0} |W(j\omega)|\): the low frequency gain
|
||||
\item \(G_\infty = lim_{\omega \to \infty} |W(j\omega)|\): the high frequency gain
|
||||
\item \(G_0 = \lim_{\omega \to 0} |W(j\omega)|\): the low frequency gain
|
||||
\item \(G_\infty = \lim_{\omega \to \infty} |W(j\omega)|\): the high frequency gain
|
||||
\item \(G_c = |W(j\omega_c)|\): the gain at a specific frequency \(\omega_c\) in \(\si{rad/s}\).
|
||||
\item \(n\): the slope between high and low frequency. It also corresponds to the order of the weighting function.
|
||||
\end{itemize}
|
||||
|
||||
The parameters \(G_0\), \(G_c\) and \(G_\infty\) should either satisfy condition \eqref{eq:cond_formula_1} or \eqref{eq:cond_formula_2}.
|
||||
The parameters \(G_0\), \(G_c\) and \(G_\infty\) should either satisfy condition~\eqref{eq:cond_formula_1} or~\eqref{eq:cond_formula_2}.
|
||||
\begin{subequations}
|
||||
\label{eq:condition_params_formula}
|
||||
\begin{align}
|
||||
@ -410,16 +400,15 @@ The parameters \(G_0\), \(G_c\) and \(G_\infty\) should either satisfy condition
|
||||
\end{align}
|
||||
\end{subequations}
|
||||
|
||||
An example of the obtained magnitude of a weighting function generated using \eqref{eq:weight_formula} is shown in Fig.~\ref{fig:weight_formula}.
|
||||
An example of the obtained magnitude of a weighting function generated using~\eqref{eq:weight_formula} is shown in Fig.~\ref{fig:weight_formula}.
|
||||
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/weight_formula.pdf}
|
||||
\caption{\label{fig:weight_formula}Magnitude of a weighting function generated using the proposed formula \eqref{eq:weight_formula}, \(G_0 = 1e^{-3}\), \(G_\infty = 10\), \(\omega_c = \SI{10}{Hz}\), \(G_c = 2\), \(n = 3\)}
|
||||
\caption{\label{fig:weight_formula}Magnitude of a weighting function generated using the proposed formula~\eqref{eq:weight_formula}, \(G_0 = 1e^{-3}\), \(G_\infty = 10\), \(\omega_c = \SI{10}{Hz}\), \(G_c = 2\), \(n = 3\).}
|
||||
\end{figure}
|
||||
|
||||
\subsection{Validation of the proposed synthesis method}
|
||||
\label{sec:org254cb4b}
|
||||
\label{sec:hinf_example}
|
||||
|
||||
The proposed methodology for the design of complementary filters is now applied on a simple example where two complementary filters \(H_1(s)\) and \(H_2(s)\) have to be designed such that:
|
||||
@ -430,7 +419,7 @@ The proposed methodology for the design of complementary filters is now applied
|
||||
\end{itemize}
|
||||
|
||||
The first step is to translate the above requirements into the design of the weighting functions.
|
||||
The proposed formula \eqref{eq:weight_formula} is here used for such purpose.
|
||||
The proposed formula~\eqref{eq:weight_formula} is here used for such purpose.
|
||||
Parameters used are summarized in Table~\ref{tab:weights_params}.
|
||||
The inverse magnitudes of the designed weighting functions, which are representing the maximum allowed norms of the complementary filters, are shown by the dashed lines in Fig.~\ref{fig:weights_W1_W2}.
|
||||
|
||||
@ -448,19 +437,19 @@ Parameters & \(W_1(s)\) & \(W_2(s)\)\\
|
||||
\(n\) & \(2\) & \(3\)\\
|
||||
\bottomrule
|
||||
\end{tabularx}
|
||||
\captionof{table}{\label{tab:weights_params}Parameters used for weighting functions \(W_1(s)\) and \(W_2(s)\) using \eqref{eq:weight_formula}}
|
||||
\captionof{table}{\label{tab:weights_params}Parameters used for weighting functions \(W_1(s)\) and \(W_2(s)\) using~\eqref{eq:weight_formula}.}
|
||||
\end{minipage}
|
||||
\hfill
|
||||
\begin{minipage}[b]{0.49\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/weights_W1_W2.pdf}
|
||||
\caption{\label{fig:weights_W1_W2}Inverse magnitude of the weighting functions}
|
||||
\caption{\label{fig:weights_W1_W2}Inverse magnitude of the weighting functions.}
|
||||
\end{minipage}
|
||||
\end{figure}
|
||||
|
||||
The \(\mathcal{H}_\infty\) synthesis is then applied to the generalized plant of Fig.~\ref{fig:h_infinity_robust_fusion_plant} on efficiently solved in Matlab~\cite{matlab20} using the Robust Control Toolbox.
|
||||
The filter \(H_2(s)\) that minimizes the \(\mathcal{H}_\infty\) norm between \(w\) and \([z_1,\ z_2]^T\) is obtained.
|
||||
The \(\mathcal{H}_\infty\) norm is here found to be close to one \eqref{eq:hinf_synthesis_result} which indicates that the synthesis is successful: the complementary filters norms are below the maximum specified upper bounds.
|
||||
The \(\mathcal{H}_\infty\) norm is here found to be close to one~\eqref{eq:hinf_synthesis_result} which indicates that the synthesis is successful: the complementary filters norms are below the maximum specified upper bounds.
|
||||
This is confirmed by the bode plots of the obtained complementary filters in Fig.~\ref{fig:hinf_filters_results}.
|
||||
|
||||
\begin{equation}
|
||||
@ -468,7 +457,7 @@ This is confirmed by the bode plots of the obtained complementary filters in Fig
|
||||
\left\|\begin{matrix} \left(1 - H_2(s)\right) W_1(s) \\ H_2(s) W_2(s) \end{matrix}\right\|_\infty \approx 1
|
||||
\end{equation}
|
||||
|
||||
Their transfer functions in the Laplace domain are given in \eqref{eq:hinf_synthesis_result_tf}.
|
||||
Their transfer functions in the Laplace domain are given in~\eqref{eq:hinf_synthesis_result_tf}.
|
||||
As expected, the obtained filters are of order \(5\), that is the sum of the weighting functions orders.
|
||||
|
||||
\begin{subequations}
|
||||
@ -483,14 +472,13 @@ As expected, the obtained filters are of order \(5\), that is the sum of the wei
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/hinf_filters_results.pdf}
|
||||
\caption{\label{fig:hinf_filters_results}Bode plot of the obtained complementary filters}
|
||||
\caption{\label{fig:hinf_filters_results}Bode plot of the obtained complementary filters.}
|
||||
\end{figure}
|
||||
|
||||
This simple example illustrates the fact that the proposed methodology for complementary filters shaping is easy to use and effective.
|
||||
A more complex real life example is taken up in the next section.
|
||||
|
||||
\section{Application: Design of Complementary Filters used in the Active Vibration Isolation System at the LIGO}
|
||||
\label{sec:org87f5cba}
|
||||
\label{sec:application_ligo}
|
||||
Sensor fusion using complementary filters are widely used in active vibration isolation systems in gravitational wave detectors such at the LIGO~\cite{matichard15_seism_isolat_advan_ligo,hua05_low_ligo}, the VIRGO~\cite{lucia18_low_frequen_optim_perfor_advan,heijningen18_low} and the KAGRA \cite[Chap. 5]{sekiguchi16_study_low_frequen_vibrat_isolat_system}.
|
||||
|
||||
@ -502,18 +490,17 @@ A simplified block diagram of the sensor fusion architecture is shown in Fig.~\r
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/ligo_super_sensor_architecture.pdf}
|
||||
\caption{\label{fig:ligo_super_sensor_architecture}Simplified block diagram of the sensor blending strategy for the first stage at the LIGO~\cite{hua04_low_ligo}}
|
||||
\caption{\label{fig:ligo_super_sensor_architecture}Simplified block diagram of the sensor blending strategy for the first stage at the LIGO~\cite{hua04_low_ligo}.}
|
||||
\end{figure}
|
||||
|
||||
The fusion of the position sensor at low frequency with the ``inertial super sensor'' at high frequency using the complementary filters (\(L_1,H_1\)) is done for several reasons, first of which is to give the super sensor a DC sensibility and therefore allow the feedback loop to have authority at zero frequency.
|
||||
The requirements on those filters are very tight and thus their design is complex and should be expressed as an optimization problem.
|
||||
The requirements on those filters are stringent and thus their design is complex and should be expressed as an optimization problem.
|
||||
|
||||
The approach used in~\cite{hua05_low_ligo} is to use FIR complementary filters and to write the synthesis as a convex optimization problem.
|
||||
After synthesis, the obtained FIR filters were found to be compliant with the requirements.
|
||||
However they are of very high order so their implementation is quite complex.
|
||||
However they are of high order so their implementation is quite complex.
|
||||
In this section, the effectiveness of the proposed complementary filter synthesis strategy is demonstrated on the same set of requirements.
|
||||
\subsection{Complementary Filters Specifications}
|
||||
\label{sec:orgf53acf5}
|
||||
\label{sec:ligo_specifications}
|
||||
|
||||
The specifications for the set of complementary filters (\(L_1,H_1\)) used at the LIGO are summarized below (for further details, refer to~\cite{hua04_polyp_fir_compl_filter_contr_system}):
|
||||
@ -530,11 +517,10 @@ They are physically represented in Fig.~\ref{fig:fir_filter_ligo} as well as the
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/fir_filter_ligo.pdf}
|
||||
\caption{\label{fig:fir_filter_ligo}Specifications and Bode plot of the obtained FIR complementary filters in~\cite{hua05_low_ligo}. The filters are here obtained using the SeDuMi Matlab toolbox~\cite{sturm99_using_sedum}}
|
||||
\caption{\label{fig:fir_filter_ligo}Specifications and Bode plot of the obtained FIR complementary filters in~\cite{hua05_low_ligo}. The filters are here obtained using the SeDuMi Matlab toolbox~\cite{sturm99_using_sedum}.}
|
||||
\end{figure}
|
||||
|
||||
\subsection{Weighting Functions Design}
|
||||
\label{sec:org4b996f7}
|
||||
\label{sec:ligo_weights}
|
||||
|
||||
The weighting functions should be designed such that their inverse magnitude is as close as possible to the specifications in order to not over-constrain the synthesis problem.
|
||||
@ -547,42 +533,39 @@ The magnitudes of the weighting functions are shown in Fig.~\ref{fig:ligo_weight
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/ligo_weights.pdf}
|
||||
\caption{\label{fig:ligo_weights}Specifications and weighting functions inverse magnitudes}
|
||||
\caption{\label{fig:ligo_weights}Specifications and weighting functions inverse magnitudes.}
|
||||
\end{figure}
|
||||
|
||||
\subsection{\(\mathcal{H}_\infty\) Synthesis of the complementary filters}
|
||||
\label{sec:org64068df}
|
||||
\label{sec:ligo_results}
|
||||
|
||||
The proposed \(\mathcal{H}_\infty\) synthesis is performed on the generalized plant shown in Fig.~\ref{fig:h_infinity_robust_fusion_plant}.
|
||||
After optimization, the \(\mathcal{H}_\infty\) norm from \(w\) to \([z_1,\ z_2]^T\) is found close to one indication successful synthesis.
|
||||
In Fig.~\ref{fig:comp_fir_ligo_hinf}, the bode plot of the obtained complementary filters are compared with the FIR filters of order 512 obtained in~\cite{hua05_low_ligo}.
|
||||
Even though the complementary filters using the \(\mathcal{H}_\infty\) synthesis are of much lower order (order 27), they are found to be very close to the FIR filters.
|
||||
Even though the complementary filters using the \(\mathcal{H}_\infty\) synthesis are of much lower order (order 27), they are found to be close to the FIR filters.
|
||||
This confirms the effectiveness of the proposed synthesis method even when the complementary filters are subject to complex requirements.
|
||||
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/comp_fir_ligo_hinf.pdf}
|
||||
\caption{\label{fig:comp_fir_ligo_hinf}Comparison of the FIR filters (dashed) designed in~\cite{hua05_low_ligo} with the filters obtained with \(\mathcal{H}_\infty\) synthesis (solid)}
|
||||
\caption{\label{fig:comp_fir_ligo_hinf}Comparison of the FIR filters (dashed) designed in~\cite{hua05_low_ligo} with the filters obtained with \(\mathcal{H}_\infty\) synthesis (solid).}
|
||||
\end{figure}
|
||||
|
||||
\section{Discussion}
|
||||
\label{sec:org36bf100}
|
||||
\label{sec:discussion}
|
||||
\subsection{``Closed-Loop'' complementary filters}
|
||||
\label{sec:org38ca6e5}
|
||||
\label{sec:closed_loop_complementary_filters}
|
||||
An alternative way to implement complementary filters is by using a fundamental property of the classical feedback architecture shown in Fig.~\ref{fig:feedback_sensor_fusion}.
|
||||
This is for instance presented in \cite{mahony05_compl_filter_desig_special_orthog,plummer06_optim_compl_filter_their_applic_motion_measur,jensen13_basic_uas}.
|
||||
This is for instance presented in~\cite{mahony05_compl_filter_desig_special_orthog,plummer06_optim_compl_filter_their_applic_motion_measur,jensen13_basic_uas}.
|
||||
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/feedback_sensor_fusion.pdf}
|
||||
\caption{\label{fig:feedback_sensor_fusion}``Closed-Loop'' complementary filters}
|
||||
\caption{\label{fig:feedback_sensor_fusion}``Closed-Loop'' complementary filters.}
|
||||
\end{figure}
|
||||
|
||||
Consider the feedback architecture of Fig.~\ref{fig:feedback_sensor_fusion}, with two inputs \(\hat{x}_1\) and \(\hat{x}_2\), and one output \(\hat{x}\).
|
||||
The output \(\hat{x}\) is linked to the inputs by \eqref{eq:closed_loop_complementary_filters}.
|
||||
The output \(\hat{x}\) is linked to the inputs by~\eqref{eq:closed_loop_complementary_filters}.
|
||||
|
||||
\begin{equation}
|
||||
\label{eq:closed_loop_complementary_filters}
|
||||
@ -602,7 +585,7 @@ Two sensors can then be merged as shown in Fig.~\ref{fig:feedback_sensor_fusion_
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/feedback_sensor_fusion_arch.pdf}
|
||||
\caption{\label{fig:feedback_sensor_fusion_arch}Classical feedback architecture used for sensor fusion}
|
||||
\caption{\label{fig:feedback_sensor_fusion_arch}Classical feedback architecture used for sensor fusion.}
|
||||
\end{figure}
|
||||
|
||||
One of the main advantage of implementing and designing complementary filters using the feedback architecture of Fig.~\ref{fig:feedback_sensor_fusion} is that all the tools of the linear control theory can be applied for the design of the filters.
|
||||
@ -610,7 +593,7 @@ If one want to shape both \(\frac{\hat{x}}{\hat{x}_1}(s) = S(s)\) and \(\frac{\h
|
||||
|
||||
To do so, weighting functions \(W_1(s)\) and \(W_2(s)\) are added to respectively shape \(S(s)\) and \(T(s)\) (Fig.~\ref{fig:feedback_synthesis_architecture}).
|
||||
Then the system is re-organized to form the generalized plant \(P_L(s)\) shown in Fig.~\ref{fig:feedback_synthesis_architecture_generalized_plant}.
|
||||
The \(\mathcal{H}_\infty\) mixed-sensitivity synthesis can finally be performed by applying the \(\mathcal{H}_\infty\) synthesis to the generalized plant \(P_L(s)\) which is described by \eqref{eq:generalized_plant_mixed_sensitivity}.
|
||||
The \(\mathcal{H}_\infty\) mixed-sensitivity synthesis can finally be performed by applying the \(\mathcal{H}_\infty\) synthesis to the generalized plant \(P_L(s)\) which is described by~\eqref{eq:generalized_plant_mixed_sensitivity}.
|
||||
|
||||
\begin{equation}
|
||||
\label{eq:generalized_plant_mixed_sensitivity}
|
||||
@ -620,7 +603,7 @@ The \(\mathcal{H}_\infty\) mixed-sensitivity synthesis can finally be performed
|
||||
\end{bmatrix}
|
||||
\end{equation}
|
||||
|
||||
The output of the synthesis is a filter \(L(s)\) such that the ``closed-loop'' \(\mathcal{H}_\infty\) norm from \([w_1,\ w_2]\) to \(z\) is less than one \eqref{eq:comp_filters_feedback_obj}.
|
||||
The output of the synthesis is a filter \(L(s)\) such that the ``closed-loop'' \(\mathcal{H}_\infty\) norm from \([w_1,\ w_2]\) to \(z\) is less than one~\eqref{eq:comp_filters_feedback_obj}.
|
||||
|
||||
\begin{equation}
|
||||
\label{eq:comp_filters_feedback_obj}
|
||||
@ -635,22 +618,22 @@ The sensor fusion can then be implemented as shown in Fig.~\ref{fig:feedback_sen
|
||||
H_1(s) = \frac{1}{1 + L(s)}; \quad H_2(s) = \frac{L(s)}{1 + L(s)}
|
||||
\end{equation}
|
||||
|
||||
The two architecture are equivalent regarding their inputs/outputs relationships.
|
||||
The two architectures are equivalent regarding their inputs/outputs relationships.
|
||||
|
||||
\begin{figure}[htbp]
|
||||
\begin{subfigure}[t]{0.6\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/feedback_synthesis_architecture.pdf}
|
||||
\caption{\label{fig:feedback_synthesis_architecture} Feedback architecture with included weights}
|
||||
\caption{\label{fig:feedback_synthesis_architecture} Feedback architecture with included weights.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\begin{subfigure}[t]{0.4\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/feedback_synthesis_architecture_generalized_plant.pdf}
|
||||
\caption{\label{fig:feedback_synthesis_architecture_generalized_plant} Generalized plant}
|
||||
\caption{\label{fig:feedback_synthesis_architecture_generalized_plant} Generalized plant.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\caption{\label{fig:h_inf_mixed_sensitivity_synthesis} $\mathcal{H}_\infty$ mixed-sensitivity synthesis}
|
||||
\caption{\label{fig:h_inf_mixed_sensitivity_synthesis} $\mathcal{H}_\infty$ mixed-sensitivity synthesis.}
|
||||
\centering
|
||||
\end{figure}
|
||||
|
||||
@ -663,11 +646,10 @@ The obtained ``closed-loop'' complementary filters are indeed equal to the ones
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/hinf_filters_results_mixed_sensitivity.pdf}
|
||||
\caption{\label{fig:hinf_filters_results_mixed_sensitivity}Bode plot of the obtained complementary filters after \(\mathcal{H}_\infty\) mixed-sensitivity synthesis}
|
||||
\caption{\label{fig:hinf_filters_results_mixed_sensitivity}Bode plot of the obtained complementary filters after \(\mathcal{H}_\infty\) mixed-sensitivity synthesis.}
|
||||
\end{figure}
|
||||
|
||||
\subsection{Synthesis of more than two Complementary Filters}
|
||||
\label{sec:org16306e2}
|
||||
\label{sec:hinf_three_comp_filters}
|
||||
|
||||
Some applications may require to merge more than two sensors~\cite{stoten01_fusion_kinet_data_using_compos_filter,becker15_compl_filter_desig_three_frequen_bands}.
|
||||
@ -684,15 +666,15 @@ A generalization of the proposed synthesis method of complementary filters is pr
|
||||
\begin{subfigure}[b]{0.59\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_fusion_three_sequential.pdf}
|
||||
\caption{\label{fig:sensor_fusion_three_sequential}Sequential fusion}
|
||||
\caption{\label{fig:sensor_fusion_three_sequential}Sequential fusion.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\begin{subfigure}[b]{0.39\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/sensor_fusion_three_parallel.pdf}
|
||||
\caption{\label{fig:sensor_fusion_three_parallel}Parallel fusion}
|
||||
\caption{\label{fig:sensor_fusion_three_parallel}Parallel fusion.}
|
||||
\end{subfigure}
|
||||
\caption{\label{fig:sensor_fusion_three}Possible sensor fusion architecture when more than two sensors are to be merged}
|
||||
\caption{\label{fig:sensor_fusion_three}Possible sensor fusion architecture when more than two sensors are to be merged.}
|
||||
\centering
|
||||
\end{figure}
|
||||
|
||||
@ -706,7 +688,7 @@ The synthesis objective is to compute a set of \(n\) stable transfer functions \
|
||||
\end{subequations}
|
||||
where \([W_1(s),\ W_2(s),\ \dots,\ W_n(s)]\) are weighting transfer functions that are chosen to specify the maximum wanted norms of the complementary filters during the synthesis.
|
||||
|
||||
Such synthesis objective is very close to the one described in Section~\ref{sec:synthesis_objective}, and indeed the proposed synthesis method is a generalization of the one presented in Section~\ref{sec:hinf_synthesis}. \par
|
||||
Such synthesis objective is closely related to the one described in Section~\ref{sec:synthesis_objective}, and indeed the proposed synthesis method is a generalization of the one presented in Section~\ref{sec:hinf_synthesis}. \par
|
||||
|
||||
Before presenting the generalized synthesis method, the case with three sensors is presented.
|
||||
Consider the generalized plant \(P_3(s)\) shown in Fig.~\ref{fig:comp_filter_three_hinf_gen_plant} which is also described by~\eqref{eq:generalized_plant_three_filters}.
|
||||
@ -720,17 +702,17 @@ Consider the generalized plant \(P_3(s)\) shown in Fig.~\ref{fig:comp_filter_thr
|
||||
\begin{subfigure}[t]{0.5\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/comp_filter_three_hinf_gen_plant.pdf}
|
||||
\caption{\label{fig:comp_filter_three_hinf_gen_plant} Generalized plant}
|
||||
\caption{\label{fig:comp_filter_three_hinf_gen_plant} Generalized plant.}
|
||||
\vfill
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\begin{subfigure}[t]{0.5\linewidth}
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/comp_filter_three_hinf_fb.pdf}
|
||||
\caption{\label{fig:comp_filter_three_hinf_fb} Generalized plant with the synthesized filter}
|
||||
\caption{\label{fig:comp_filter_three_hinf_fb} Generalized plant with the synthesized filter.}
|
||||
\end{subfigure}
|
||||
\hfill
|
||||
\caption{\label{fig:comp_filter_three_hinf} Architecture for the $\mathcal{H}_\infty$ synthesis of three complementary filters}
|
||||
\caption{\label{fig:comp_filter_three_hinf} Architecture for the $\mathcal{H}_\infty$ synthesis of three complementary filters.}
|
||||
\centering
|
||||
\end{figure}
|
||||
|
||||
@ -741,7 +723,7 @@ Applying the \(\mathcal{H}_\infty\) synthesis on the generalized plant \(P_3(s)\
|
||||
\left\| \begin{matrix} \left[1 - H_2(s) - H_3(s)\right] W_1(s) \\ H_2(s) W_2(s) \\ H_3(s) W_3(s) \end{matrix} \right\|_\infty \le 1
|
||||
\end{equation}
|
||||
|
||||
By defining \(H_1(s) \triangleq 1 - H_2(s) - H_3(s)\), the proposed \(\mathcal{H}_\infty\) synthesis solves the design problem \eqref{eq:hinf_problem_gen} with \(n=3\). \par
|
||||
By defining \(H_1(s) \triangleq 1 - H_2(s) - H_3(s)\), the proposed \(\mathcal{H}_\infty\) synthesis solves the design problem~\eqref{eq:hinf_problem_gen} with \(n=3\). \par
|
||||
|
||||
An example is given to validate the method where three sensors are used in different frequency bands.
|
||||
For instance a displacement sensor from DC up to \(\SI{1}{Hz}\), a geophone from \(1\) to \(\SI{10}{Hz}\) and an accelerometer above \(\SI{10}{Hz}\).
|
||||
@ -751,7 +733,7 @@ The \(\mathcal{H}_\infty\) synthesis is performed on the generalized plant \(P_3
|
||||
\begin{figure}[htbp]
|
||||
\centering
|
||||
\includegraphics[scale=1]{figs/three_complementary_filters_results.pdf}
|
||||
\caption{\label{fig:three_complementary_filters_results}Bode plot of the inverse weighting functions and of the three complementary filters obtained using the \(\mathcal{H}_\infty\) synthesis}
|
||||
\caption{\label{fig:three_complementary_filters_results}Bode plot of the inverse weighting functions and of the three complementary filters obtained using the \(\mathcal{H}_\infty\) synthesis.}
|
||||
\end{figure}
|
||||
|
||||
Even though there might not be any practical application for a set of more than 3 complementary filters, it can still be designed using the same procedure.
|
||||
@ -771,27 +753,20 @@ A set of \(n\) complementary filters can be shaped using the generalized plant \
|
||||
\end{equation}
|
||||
|
||||
\section{Conclusion}
|
||||
\label{sec:org05380ec}
|
||||
\label{sec:conclusion}
|
||||
|
||||
Sensors measuring a physical quantities are always subject to limitations both in terms of bandwidth or accuracy.
|
||||
Complementary filters can be used to fuse multiple sensors with different characteristics in order to combine their benefits and yield a better estimate of the measured physical quantity.
|
||||
|
||||
The sensor fusion robustness and the obtained super sensor noise can be linked to the magnitude of the complementary filters.
|
||||
In this paper, a synthesis method that enables the shaping of the complementary filters norms has been proposed.
|
||||
Several example were used to emphasize the simplicity and the effectiveness of the proposed method.
|
||||
Links with ``closed-loop'' complementary filters where highlighted, and the proposed method was generalized for the design of a set of more than two complementary filters.
|
||||
Several examples were used to emphasize the simplicity and the effectiveness of the proposed method.
|
||||
Links with ``closed-loop'' complementary filters were highlighted, and the proposed method was generalized for designing a set of more than two complementary filters.
|
||||
|
||||
Future work will aim at developing a complementary filter synthesis method that minimizes the super sensor noise while ensuring the robustness of the fusion.
|
||||
The future work will aim at developing a complementary filter synthesis method that minimizes the super sensor noise while ensuring the robustness of the fusion.
|
||||
|
||||
\section*{Acknowledgment}
|
||||
\label{sec:orgcfbe3a3}
|
||||
This research benefited from a FRIA grant from the French Community of Belgium.
|
||||
This paper is based on a paper previously presented at the ICCMA conference~\cite{dehaeze19_compl_filter_shapin_using_synth}.
|
||||
|
||||
\section*{Data Availability}
|
||||
\label{sec:orga0214e5}
|
||||
Matlab \cite{matlab20} was used for this study.
|
||||
Matlab~\cite{matlab20} was used for this study.
|
||||
The source code is available under a MIT License and archived in Zenodo~\cite{dehaeze21_new_method_desig_compl_filter_code}.
|
||||
|
||||
\bibliographystyle{elsarticle-num}
|
||||
|
||||
@ -3,9 +3,9 @@
|
||||
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
|
||||
<html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="en">
|
||||
<head>
|
||||
<!-- 2021-09-01 mer. 16:59 -->
|
||||
<!-- 2021-09-02 jeu. 10:04 -->
|
||||
<meta http-equiv="Content-Type" content="text/html;charset=utf-8" />
|
||||
<title>A new method of designing complementary filters for sensor fusion using the $\mathcal{H}_\infty$ synthesis - Matlab Computation</title>
|
||||
<title>Designing Complementary filters for sensor fusion using $\mathcal{H}_\infty$ synthesis - Matlab Computation</title>
|
||||
<meta name="author" content="Dehaeze Thomas" />
|
||||
<meta name="generator" content="Org Mode" />
|
||||
<link rel="stylesheet" type="text/css" href="https://research.tdehaeze.xyz/css/style.css"/>
|
||||
@ -34,42 +34,42 @@
|
||||
|
|
||||
<a accesskey="H" href="../index.html"> HOME </a>
|
||||
</div><div id="content" class="content">
|
||||
<h1 class="title">A new method of designing complementary filters for sensor fusion using the \(\mathcal{H}_\infty\) synthesis - Matlab Computation</h1>
|
||||
<h1 class="title">Designing Complementary filters for sensor fusion using \(\mathcal{H}_\infty\) synthesis - Matlab Computation</h1>
|
||||
<div id="table-of-contents" role="doc-toc">
|
||||
<h2>Table of Contents</h2>
|
||||
<div id="text-table-of-contents" role="doc-toc">
|
||||
<ul>
|
||||
<li><a href="#sec:h_inf_synthesis_complementary_filters">1. H-Infinity synthesis of complementary filters</a>
|
||||
<ul>
|
||||
<li><a href="#org039cf66">1.1. Synthesis Architecture</a></li>
|
||||
<li><a href="#org5504635">1.2. Design of Weighting Function - Proposed formula</a></li>
|
||||
<li><a href="#orgbe5ea75">1.3. Weighting functions for the design of two complementary filters</a></li>
|
||||
<li><a href="#org93a8bd3">1.4. Synthesis of the complementary filters</a></li>
|
||||
<li><a href="#orgd4544b8">1.5. Obtained Complementary Filters</a></li>
|
||||
<li><a href="#orgc4b4c6d">1.1. Synthesis Architecture</a></li>
|
||||
<li><a href="#org99141ab">1.2. Design of Weighting Function - Proposed formula</a></li>
|
||||
<li><a href="#orge755bd5">1.3. Weighting functions for the design of two complementary filters</a></li>
|
||||
<li><a href="#orgbd5024b">1.4. Synthesis of the complementary filters</a></li>
|
||||
<li><a href="#org8eadcab">1.5. Obtained Complementary Filters</a></li>
|
||||
</ul>
|
||||
</li>
|
||||
<li><a href="#sec:comp_filters_ligo">2. Design of complementary filters used in the Active Vibration Isolation System at the LIGO</a>
|
||||
<ul>
|
||||
<li><a href="#org0ef35df">2.1. Specifications</a></li>
|
||||
<li><a href="#org86f81f8">2.2. FIR Filter</a></li>
|
||||
<li><a href="#orga3462b9">2.3. Weighting function design</a></li>
|
||||
<li><a href="#org2d35692">2.4. Synthesis of the complementary filters</a></li>
|
||||
<li><a href="#org2d8f858">2.5. Comparison of the FIR filters and synthesized filters</a></li>
|
||||
<li><a href="#org775cee5">2.1. Specifications</a></li>
|
||||
<li><a href="#org6402809">2.2. FIR Filter</a></li>
|
||||
<li><a href="#orge47e4d1">2.3. Weighting function design</a></li>
|
||||
<li><a href="#org5ef288a">2.4. Synthesis of the complementary filters</a></li>
|
||||
<li><a href="#org71076fb">2.5. Comparison of the FIR filters and synthesized filters</a></li>
|
||||
</ul>
|
||||
</li>
|
||||
<li><a href="#sec:closed_loop_complementary_filters">3. “Closed-Loop” complementary filters</a>
|
||||
<ul>
|
||||
<li><a href="#orgafe7677">3.1. Weighting Function design</a></li>
|
||||
<li><a href="#org746a218">3.2. Generalized plant</a></li>
|
||||
<li><a href="#org18bfdfc">3.3. Synthesis of the closed-loop complementary filters</a></li>
|
||||
<li><a href="#orge450acb">3.4. Synthesized filters</a></li>
|
||||
<li><a href="#orgfa47886">3.1. Weighting Function design</a></li>
|
||||
<li><a href="#org43ccb7c">3.2. Generalized plant</a></li>
|
||||
<li><a href="#orgbf556a9">3.3. Synthesis of the closed-loop complementary filters</a></li>
|
||||
<li><a href="#org426fd90">3.4. Synthesized filters</a></li>
|
||||
</ul>
|
||||
</li>
|
||||
<li><a href="#sec:three_comp_filters">4. Synthesis of three complementary filters</a>
|
||||
<ul>
|
||||
<li><a href="#orga9cd2fd">4.1. Synthesis Architecture</a></li>
|
||||
<li><a href="#orged7fb64">4.2. Weights</a></li>
|
||||
<li><a href="#org0eb6887">4.3. H-Infinity Synthesis</a></li>
|
||||
<li><a href="#org23dc1a0">4.1. Synthesis Architecture</a></li>
|
||||
<li><a href="#orgf36e7cb">4.2. Weights</a></li>
|
||||
<li><a href="#org8bb1bfb">4.3. H-Infinity Synthesis</a></li>
|
||||
</ul>
|
||||
</li>
|
||||
<li><a href="#sec:matlab_scripts">5. Matlab Scripts</a>
|
||||
@ -110,27 +110,27 @@ This document is divided into the following sections also corresponding to the p
|
||||
<div class="outline-text-2" id="text-sec:h_inf_synthesis_complementary_filters">
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org039cf66" class="outline-3">
|
||||
<h3 id="org039cf66"><span class="section-number-3">1.1.</span> Synthesis Architecture</h3>
|
||||
<div id="outline-container-orgc4b4c6d" class="outline-3">
|
||||
<h3 id="orgc4b4c6d"><span class="section-number-3">1.1.</span> Synthesis Architecture</h3>
|
||||
<div class="outline-text-3" id="text-1-1">
|
||||
<p>
|
||||
In order to generate two complementary filters with a wanted shape, the generalized plant of Figure <a href="#org8ec6b4a">1</a> can be used.
|
||||
In order to generate two complementary filters with a wanted shape, the generalized plant of Figure <a href="#org938abd0">1</a> can be used.
|
||||
The included weights \(W_1(s)\) and \(W_2(s)\) are used to specify the upper bounds of the complementary filters being generated.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="org8ec6b4a" class="figure">
|
||||
<div id="org938abd0" class="figure">
|
||||
<p><img src="figs-journal/h_infinity_robust_fusion_plant.png" alt="h_infinity_robust_fusion_plant.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 1: </span>Generalized plant used for the \(\mathcal{H}_\infty\) synthesis of a set of two complementary fiters</p>
|
||||
</div>
|
||||
|
||||
<p>
|
||||
Applied the standard \(\mathcal{H}_\infty\) synthesis on this generalized plant will give a transfer function \(H_2(s)\) (see Figure <a href="#org38fe996">2</a>) such that the \(\mathcal{H}_\infty\) norm of the transfer function from \(w\) to \([z_1,\ z_2]\) is less than one \eqref{eq:h_inf_objective}.
|
||||
Applied the standard \(\mathcal{H}_\infty\) synthesis on this generalized plant will give a transfer function \(H_2(s)\) (see Figure <a href="#org133e437">2</a>) such that the \(\mathcal{H}_\infty\) norm of the transfer function from \(w\) to \([z_1,\ z_2]\) is less than one \eqref{eq:h_inf_objective}.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="org38fe996" class="figure">
|
||||
<div id="org133e437" class="figure">
|
||||
<p><img src="figs-journal/h_infinity_robust_fusion_fb.png" alt="h_infinity_robust_fusion_fb.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 2: </span>Generalized plant with the synthesized filter obtained after the \(\mathcal{H}_\infty\) synthesis</p>
|
||||
@ -173,8 +173,8 @@ The complete Matlab script for this part is given in Section <a href="#sec:1_syn
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org5504635" class="outline-3">
|
||||
<h3 id="org5504635"><span class="section-number-3">1.2.</span> Design of Weighting Function - Proposed formula</h3>
|
||||
<div id="outline-container-org99141ab" class="outline-3">
|
||||
<h3 id="org99141ab"><span class="section-number-3">1.2.</span> Design of Weighting Function - Proposed formula</h3>
|
||||
<div class="outline-text-3" id="text-1-2">
|
||||
<p>
|
||||
A formula is proposed to help the design of the weighting functions:
|
||||
@ -198,11 +198,11 @@ The parameters permits to specify:
|
||||
</ul>
|
||||
|
||||
<p>
|
||||
The general shape of a weighting function generated using the formula is shown in figure <a href="#orga5d226a">3</a>.
|
||||
The general shape of a weighting function generated using the formula is shown in figure <a href="#org652005a">3</a>.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="orga5d226a" class="figure">
|
||||
<div id="org652005a" class="figure">
|
||||
<p><img src="figs/weight_formula.png" alt="weight_formula.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 3: </span>Magnitude of the weighting function generated using formula \eqref{eq:weighting_function_formula}</p>
|
||||
@ -210,11 +210,11 @@ The general shape of a weighting function generated using the formula is shown i
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orgbe5ea75" class="outline-3">
|
||||
<h3 id="orgbe5ea75"><span class="section-number-3">1.3.</span> Weighting functions for the design of two complementary filters</h3>
|
||||
<div id="outline-container-orge755bd5" class="outline-3">
|
||||
<h3 id="orge755bd5"><span class="section-number-3">1.3.</span> Weighting functions for the design of two complementary filters</h3>
|
||||
<div class="outline-text-3" id="text-1-3">
|
||||
<p>
|
||||
<a id="org3789fe5"></a>
|
||||
<a id="orgbf8ba5a"></a>
|
||||
</p>
|
||||
|
||||
<p>
|
||||
@ -233,11 +233,11 @@ W2 = generateWF(<span class="org-string">'n'</span>, 2, <span class="org-string"
|
||||
</div>
|
||||
|
||||
<p>
|
||||
The inverse magnitude of these two weighting functions are shown in Figure <a href="#org41b9a7a">4</a>.
|
||||
The inverse magnitude of these two weighting functions are shown in Figure <a href="#org9dd5099">4</a>.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="org41b9a7a" class="figure">
|
||||
<div id="org9dd5099" class="figure">
|
||||
<p><img src="figs/weights_W1_W2.png" alt="weights_W1_W2.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 4: </span>Inverse magnitude of the design weighting functions</p>
|
||||
@ -245,11 +245,11 @@ The inverse magnitude of these two weighting functions are shown in Figure <a hr
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org93a8bd3" class="outline-3">
|
||||
<h3 id="org93a8bd3"><span class="section-number-3">1.4.</span> Synthesis of the complementary filters</h3>
|
||||
<div id="outline-container-orgbd5024b" class="outline-3">
|
||||
<h3 id="orgbd5024b"><span class="section-number-3">1.4.</span> Synthesis of the complementary filters</h3>
|
||||
<div class="outline-text-3" id="text-1-4">
|
||||
<p>
|
||||
The generalized plant of Figure <a href="#org8ec6b4a">1</a> is defined as follows:
|
||||
The generalized plant of Figure <a href="#org938abd0">1</a> is defined as follows:
|
||||
</p>
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-matlab"><span class="org-matlab-cellbreak">%% Generalized Plant</span>
|
||||
@ -268,7 +268,7 @@ And the \(\mathcal{H}_\infty\) synthesis is performed using the <code>hinfsyn</c
|
||||
</pre>
|
||||
</div>
|
||||
|
||||
<pre class="example" id="orgdbfdcce">
|
||||
<pre class="example" id="orge151ac9">
|
||||
[H2, ~, gamma, ~] = hinfsyn(P, 1, 1,'TOLGAM', 0.001, 'METHOD', 'ric', 'DISPLAY', 'on');
|
||||
|
||||
Test bounds: 0.3223 <= gamma <= 1000
|
||||
@ -318,15 +318,15 @@ This function is described in Section <a href="#sec:generateCF">5.6</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orgd4544b8" class="outline-3">
|
||||
<h3 id="orgd4544b8"><span class="section-number-3">1.5.</span> Obtained Complementary Filters</h3>
|
||||
<div id="outline-container-org8eadcab" class="outline-3">
|
||||
<h3 id="org8eadcab"><span class="section-number-3">1.5.</span> Obtained Complementary Filters</h3>
|
||||
<div class="outline-text-3" id="text-1-5">
|
||||
<p>
|
||||
The obtained complementary filters are shown below and are found to be of order 5.
|
||||
Their bode plots are shown in figure <a href="#org1c88e24">5</a> and compare with the defined upper bounds.
|
||||
Their bode plots are shown in figure <a href="#org2c627be">5</a> and compare with the defined upper bounds.
|
||||
</p>
|
||||
|
||||
<pre class="example" id="org6f67900">
|
||||
<pre class="example" id="orgfedb8df">
|
||||
zpk(H1)
|
||||
ans =
|
||||
|
||||
@ -343,7 +343,7 @@ ans =
|
||||
</pre>
|
||||
|
||||
|
||||
<div id="org1c88e24" class="figure">
|
||||
<div id="org2c627be" class="figure">
|
||||
<p><img src="figs/hinf_filters_results.png" alt="hinf_filters_results.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 5: </span>Obtained complementary filters using \(\mathcal{H}_\infty\) synthesis</p>
|
||||
@ -364,8 +364,8 @@ The complete Matlab script for this part is given in Section <a href="#sec:2_lig
|
||||
</p>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org0ef35df" class="outline-3">
|
||||
<h3 id="org0ef35df"><span class="section-number-3">2.1.</span> Specifications</h3>
|
||||
<div id="outline-container-org775cee5" class="outline-3">
|
||||
<h3 id="org775cee5"><span class="section-number-3">2.1.</span> Specifications</h3>
|
||||
<div class="outline-text-3" id="text-2-1">
|
||||
<p>
|
||||
The specifications for the set of complementary filters (\(L_1,H_1\)) used at the LIGO are summarized below (for further details, refer to (<a href="#citeproc_bib_item_4">Hua, Debra, et al. 2004</a>)):
|
||||
@ -378,11 +378,11 @@ The specifications for the set of complementary filters (\(L_1,H_1\)) used at th
|
||||
</ul>
|
||||
|
||||
<p>
|
||||
The specifications are translated into upper bounds of the complementary filters and are shown in Figure <a href="#org4e806e6">6</a>.
|
||||
The specifications are translated into upper bounds of the complementary filters and are shown in Figure <a href="#org03607c6">6</a>.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="org4e806e6" class="figure">
|
||||
<div id="org03607c6" class="figure">
|
||||
<p><img src="figs/ligo_specifications.png" alt="ligo_specifications.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 6: </span>Specification for the LIGO complementary filters</p>
|
||||
@ -390,8 +390,8 @@ The specifications are translated into upper bounds of the complementary filters
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org86f81f8" class="outline-3">
|
||||
<h3 id="org86f81f8"><span class="section-number-3">2.2.</span> FIR Filter</h3>
|
||||
<div id="outline-container-org6402809" class="outline-3">
|
||||
<h3 id="org6402809"><span class="section-number-3">2.2.</span> FIR Filter</h3>
|
||||
<div class="outline-text-3" id="text-2-2">
|
||||
<p>
|
||||
To replicated the complementary filters developed in (<a href="#citeproc_bib_item_3">Hua, Adhikari, et al. 2004</a>), the CVX Matlab toolbox (<a href="#citeproc_bib_item_2">Grant and Boyd 2014</a>) is used.
|
||||
@ -476,7 +476,7 @@ h = y(2<span class="org-builtin">:</span>end);
|
||||
</pre>
|
||||
</div>
|
||||
|
||||
<pre class="example" id="org8c893b5">
|
||||
<pre class="example" id="orge5b954e">
|
||||
cvx_begin
|
||||
variable y(n+1,1)
|
||||
% t
|
||||
@ -553,7 +553,7 @@ h = y(2:end);
|
||||
</pre>
|
||||
|
||||
<p>
|
||||
Finally, the filter response is computed over the frequency vector defined and the result is shown on figure <a href="#org4fd297c">7</a> which is very close to the filters obtain in (<a href="#citeproc_bib_item_3">Hua, Adhikari, et al. 2004</a>).
|
||||
Finally, the filter response is computed over the frequency vector defined and the result is shown on figure <a href="#orgf5ce15e">7</a> which is very close to the filters obtain in (<a href="#citeproc_bib_item_3">Hua, Adhikari, et al. 2004</a>).
|
||||
</p>
|
||||
|
||||
<div class="org-src-container">
|
||||
@ -564,7 +564,7 @@ H = [exp(<span class="org-builtin">-</span><span class="org-matlab-math">j</span
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org4fd297c" class="figure">
|
||||
<div id="orgf5ce15e" class="figure">
|
||||
<p><img src="figs/fir_filter_ligo.png" alt="fir_filter_ligo.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 7: </span>FIR Complementary filters obtain after convex optimization</p>
|
||||
@ -572,8 +572,8 @@ H = [exp(<span class="org-builtin">-</span><span class="org-matlab-math">j</span
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orga3462b9" class="outline-3">
|
||||
<h3 id="orga3462b9"><span class="section-number-3">2.3.</span> Weighting function design</h3>
|
||||
<div id="outline-container-orge47e4d1" class="outline-3">
|
||||
<h3 id="orge47e4d1"><span class="section-number-3">2.3.</span> Weighting function design</h3>
|
||||
<div class="outline-text-3" id="text-2-3">
|
||||
<p>
|
||||
The weightings function that will be used for the \(\mathcal{H}_\infty\) synthesis of the complementary filters are now designed.
|
||||
@ -635,11 +635,11 @@ wL = minreal(ss(wL));
|
||||
</div>
|
||||
|
||||
<p>
|
||||
The inverse magnitude of the weighting functions are shown in Figure <a href="#orgf2af1c4">8</a>.
|
||||
The inverse magnitude of the weighting functions are shown in Figure <a href="#org588a935">8</a>.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="orgf2af1c4" class="figure">
|
||||
<div id="org588a935" class="figure">
|
||||
<p><img src="figs/ligo_weights.png" alt="ligo_weights.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 8: </span>Weights for the \(\mathcal{H}_\infty\) synthesis</p>
|
||||
@ -647,11 +647,11 @@ The inverse magnitude of the weighting functions are shown in Figure <a href="#o
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org2d35692" class="outline-3">
|
||||
<h3 id="org2d35692"><span class="section-number-3">2.4.</span> Synthesis of the complementary filters</h3>
|
||||
<div id="outline-container-org5ef288a" class="outline-3">
|
||||
<h3 id="org5ef288a"><span class="section-number-3">2.4.</span> Synthesis of the complementary filters</h3>
|
||||
<div class="outline-text-3" id="text-2-4">
|
||||
<p>
|
||||
The generalized plant of figure <a href="#org8ec6b4a">1</a> is defined.
|
||||
The generalized plant of figure <a href="#org938abd0">1</a> is defined.
|
||||
</p>
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-matlab"><span class="org-matlab-cellbreak">%% Generalized plant for the H-infinity Synthesis</span>
|
||||
@ -670,7 +670,7 @@ And the standard \(\mathcal{H}_\infty\) synthesis using the <code>hinfsyn</code>
|
||||
</pre>
|
||||
</div>
|
||||
|
||||
<pre class="example" id="orgb3aafc2">
|
||||
<pre class="example" id="org12089e0">
|
||||
[Hl, ~, gamma, ~] = hinfsyn(P, 1, 1,'TOLGAM', 0.001, 'METHOD', 'ric', 'DISPLAY', 'on');
|
||||
Resetting value of Gamma min based on D_11, D_12, D_21 terms
|
||||
|
||||
@ -714,18 +714,18 @@ Hh = 1 <span class="org-builtin">-</span> Hl;
|
||||
The size of the filters is shown to be equal to the sum of the weighting functions orders.
|
||||
</p>
|
||||
|
||||
<pre class="example" id="orga490b22">
|
||||
<pre class="example" id="org7e59413">
|
||||
size(Hh), size(Hl)
|
||||
State-space model with 1 outputs, 1 inputs, and 27 states.
|
||||
State-space model with 1 outputs, 1 inputs, and 27 states.
|
||||
</pre>
|
||||
|
||||
<p>
|
||||
The magnitude of the obtained filters as well as the requirements are shown in Figure <a href="#orgb24a37f">9</a>.
|
||||
The magnitude of the obtained filters as well as the requirements are shown in Figure <a href="#orgb9b3d25">9</a>.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="orgb24a37f" class="figure">
|
||||
<div id="orgb9b3d25" class="figure">
|
||||
<p><img src="figs/hinf_synthesis_ligo_results.png" alt="hinf_synthesis_ligo_results.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 9: </span>Obtained complementary filters using the \(\mathcal{H}_\infty\) synthesis</p>
|
||||
@ -733,19 +733,19 @@ The magnitude of the obtained filters as well as the requirements are shown in F
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org2d8f858" class="outline-3">
|
||||
<h3 id="org2d8f858"><span class="section-number-3">2.5.</span> Comparison of the FIR filters and synthesized filters</h3>
|
||||
<div id="outline-container-org71076fb" class="outline-3">
|
||||
<h3 id="org71076fb"><span class="section-number-3">2.5.</span> Comparison of the FIR filters and synthesized filters</h3>
|
||||
<div class="outline-text-3" id="text-2-5">
|
||||
<p>
|
||||
Let’s now compare the FIR filters designed in (<a href="#citeproc_bib_item_3">Hua, Adhikari, et al. 2004</a>) with the with complementary filters obtained with the \(\mathcal{H}_\infty\) synthesis.
|
||||
</p>
|
||||
|
||||
<p>
|
||||
This is done in Figure <a href="#org8f17c7a">10</a>, and both set of filters are found to be very close to each other.
|
||||
This is done in Figure <a href="#org7eeadda">10</a>, and both set of filters are found to be very close to each other.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="org8f17c7a" class="figure">
|
||||
<div id="org7eeadda" class="figure">
|
||||
<p><img src="figs/comp_fir_ligo_hinf.png" alt="comp_fir_ligo_hinf.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 10: </span>Comparison between the FIR filters developped for LIGO and the \(\mathcal{H}_\infty\) complementary filters</p>
|
||||
@ -758,11 +758,11 @@ This is done in Figure <a href="#org8f17c7a">10</a>, and both set of filters are
|
||||
<h2 id="sec:closed_loop_complementary_filters"><span class="section-number-2">3.</span> “Closed-Loop” complementary filters</h2>
|
||||
<div class="outline-text-2" id="text-sec:closed_loop_complementary_filters">
|
||||
<p>
|
||||
In this section, the classical feedback architecture shown in Figure <a href="#org915c7e1">11</a> is used for the design of complementary filters.
|
||||
In this section, the classical feedback architecture shown in Figure <a href="#org91711eb">11</a> is used for the design of complementary filters.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="org915c7e1" class="figure">
|
||||
<div id="org91711eb" class="figure">
|
||||
<p><img src="figs-journal/feedback_sensor_fusion.png" alt="feedback_sensor_fusion.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 11: </span>“Closed-Loop” complementary filters</p>
|
||||
@ -773,12 +773,12 @@ The complete Matlab script for this part is given in Section <a href="#sec:3_clo
|
||||
</p>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orgafe7677" class="outline-3">
|
||||
<h3 id="orgafe7677"><span class="section-number-3">3.1.</span> Weighting Function design</h3>
|
||||
<div id="outline-container-orgfa47886" class="outline-3">
|
||||
<h3 id="orgfa47886"><span class="section-number-3">3.1.</span> Weighting Function design</h3>
|
||||
<div class="outline-text-3" id="text-3-1">
|
||||
<p>
|
||||
Weighting functions using the <code>generateWF</code> Matlab function are designed to specify the upper bounds of the complementary filters to be designed.
|
||||
These weighting functions are the same as the ones used in Section <a href="#org3789fe5">1.3</a>.
|
||||
These weighting functions are the same as the ones used in Section <a href="#orgbf8ba5a">1.3</a>.
|
||||
</p>
|
||||
|
||||
<div class="org-src-container">
|
||||
@ -790,11 +790,11 @@ W2 = generateWF(<span class="org-string">'n'</span>, 2, <span class="org-string"
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org746a218" class="outline-3">
|
||||
<h3 id="org746a218"><span class="section-number-3">3.2.</span> Generalized plant</h3>
|
||||
<div id="outline-container-org43ccb7c" class="outline-3">
|
||||
<h3 id="org43ccb7c"><span class="section-number-3">3.2.</span> Generalized plant</h3>
|
||||
<div class="outline-text-3" id="text-3-2">
|
||||
<p>
|
||||
The generalized plant of Figure <a href="#org08ff8ac">12</a> is defined below:
|
||||
The generalized plant of Figure <a href="#org3a36382">12</a> is defined below:
|
||||
</p>
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-matlab"><span class="org-matlab-cellbreak">%% Generalized plant for "closed-loop" complementary filter synthesis</span>
|
||||
@ -804,7 +804,7 @@ P = [ W1 0 1;
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org08ff8ac" class="figure">
|
||||
<div id="org3a36382" class="figure">
|
||||
<p><img src="figs-journal/feedback_synthesis_architecture_generalized_plant.png" alt="feedback_synthesis_architecture_generalized_plant.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 12: </span>Generalized plant used for the \(\mathcal{H}_\infty\) synthesis of “closed-loop” complementary filters</p>
|
||||
@ -812,8 +812,8 @@ P = [ W1 0 1;
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org18bfdfc" class="outline-3">
|
||||
<h3 id="org18bfdfc"><span class="section-number-3">3.3.</span> Synthesis of the closed-loop complementary filters</h3>
|
||||
<div id="outline-container-orgbf556a9" class="outline-3">
|
||||
<h3 id="orgbf556a9"><span class="section-number-3">3.3.</span> Synthesis of the closed-loop complementary filters</h3>
|
||||
<div class="outline-text-3" id="text-3-3">
|
||||
<p>
|
||||
And the standard \(\mathcal{H}_\infty\) synthesis is performed.
|
||||
@ -824,7 +824,7 @@ And the standard \(\mathcal{H}_\infty\) synthesis is performed.
|
||||
</pre>
|
||||
</div>
|
||||
|
||||
<pre class="example" id="org2e786a3">
|
||||
<pre class="example" id="org1af6c18">
|
||||
Test bounds: 0.3191 <= gamma <= 1.669
|
||||
|
||||
gamma X>=0 Y>=0 rho(XY)<1 p/f
|
||||
@ -848,11 +848,11 @@ And the standard \(\mathcal{H}_\infty\) synthesis is performed.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orge450acb" class="outline-3">
|
||||
<h3 id="orge450acb"><span class="section-number-3">3.4.</span> Synthesized filters</h3>
|
||||
<div id="outline-container-org426fd90" class="outline-3">
|
||||
<h3 id="org426fd90"><span class="section-number-3">3.4.</span> Synthesized filters</h3>
|
||||
<div class="outline-text-3" id="text-3-4">
|
||||
<p>
|
||||
The obtained filter \(L(s)\) can then be included in the feedback architecture shown in Figure <a href="#orgf151405">13</a>.
|
||||
The obtained filter \(L(s)\) can then be included in the feedback architecture shown in Figure <a href="#org9d40494">13</a>.
|
||||
</p>
|
||||
|
||||
<p>
|
||||
@ -865,7 +865,7 @@ H2 = 1 <span class="org-builtin">-</span> H1;
|
||||
</pre>
|
||||
</div>
|
||||
|
||||
<pre class="example" id="orgae811fb">
|
||||
<pre class="example" id="org8cae256">
|
||||
zpk(H1) =
|
||||
(s+3.842)^3 (s+153.6) (s+1.289e05)
|
||||
-------------------------------------------------------
|
||||
@ -878,11 +878,11 @@ zpk(H2) =
|
||||
</pre>
|
||||
|
||||
<p>
|
||||
The bode plots of the synthesized complementary filters are compared with the upper bounds in Figure <a href="#orgf151405">13</a>.
|
||||
The bode plots of the synthesized complementary filters are compared with the upper bounds in Figure <a href="#org9d40494">13</a>.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="orgf151405" class="figure">
|
||||
<div id="org9d40494" class="figure">
|
||||
<p><img src="figs/hinf_filters_results_mixed_sensitivity.png" alt="hinf_filters_results_mixed_sensitivity.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 13: </span>Bode plot of the obtained complementary filters</p>
|
||||
@ -903,8 +903,8 @@ The complete Matlab script for this part is given in Section <a href="#sec:4_thr
|
||||
</p>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orga9cd2fd" class="outline-3">
|
||||
<h3 id="orga9cd2fd"><span class="section-number-3">4.1.</span> Synthesis Architecture</h3>
|
||||
<div id="outline-container-org23dc1a0" class="outline-3">
|
||||
<h3 id="org23dc1a0"><span class="section-number-3">4.1.</span> Synthesis Architecture</h3>
|
||||
<div class="outline-text-3" id="text-4-1">
|
||||
<p>
|
||||
The synthesis objective is to shape three filters that are complementary.
|
||||
@ -921,18 +921,18 @@ This corresponds to the conditions \eqref{eq:obj_three_cf} where \(W_1(s)\), \(W
|
||||
\end{equation}
|
||||
|
||||
<p>
|
||||
This synthesis can be done by performing the standard \(\mathcal{H}_\infty\) synthesis with on the generalized plant in Figure <a href="#org452ccad">14</a>.
|
||||
This synthesis can be done by performing the standard \(\mathcal{H}_\infty\) synthesis with on the generalized plant in Figure <a href="#orge2e8c55">14</a>.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="org452ccad" class="figure">
|
||||
<div id="orge2e8c55" class="figure">
|
||||
<p><img src="figs-journal/comp_filter_three_hinf_fb.png" alt="comp_filter_three_hinf_fb.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 14: </span>Generalized architecture for generating 3 complementary filters</p>
|
||||
</div>
|
||||
|
||||
<p>
|
||||
After synthesis, filter \(H_2(s)\) and \(H_3(s)\) are obtained as shown in Figure <a href="#org452ccad">14</a>.
|
||||
After synthesis, filter \(H_2(s)\) and \(H_3(s)\) are obtained as shown in Figure <a href="#orge2e8c55">14</a>.
|
||||
The last filter \(H_1(s)\) is defined as the complementary of the two others as in \eqref{eq:H1_complementary_of_H2_H3}.
|
||||
</p>
|
||||
|
||||
@ -942,8 +942,8 @@ The last filter \(H_1(s)\) is defined as the complementary of the two others as
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orged7fb64" class="outline-3">
|
||||
<h3 id="orged7fb64"><span class="section-number-3">4.2.</span> Weights</h3>
|
||||
<div id="outline-container-orgf36e7cb" class="outline-3">
|
||||
<h3 id="orgf36e7cb"><span class="section-number-3">4.2.</span> Weights</h3>
|
||||
<div class="outline-text-3" id="text-4-2">
|
||||
<p>
|
||||
The three weighting functions are defined as shown below.
|
||||
@ -958,11 +958,11 @@ W3 = generateWF(<span class="org-string">'n'</span>, 3, <span class="org-string"
|
||||
</div>
|
||||
|
||||
<p>
|
||||
Their inverse magnitudes are displayed in Figure <a href="#orge40038d">15</a>.
|
||||
Their inverse magnitudes are displayed in Figure <a href="#orga89c5b8">15</a>.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="orge40038d" class="figure">
|
||||
<div id="orga89c5b8" class="figure">
|
||||
<p><img src="figs/three_weighting_functions.png" alt="three_weighting_functions.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 15: </span>Three weighting functions used for the \(\mathcal{H}_\infty\) synthesis of the complementary filters</p>
|
||||
@ -970,11 +970,11 @@ Their inverse magnitudes are displayed in Figure <a href="#orge40038d">15</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org0eb6887" class="outline-3">
|
||||
<h3 id="org0eb6887"><span class="section-number-3">4.3.</span> H-Infinity Synthesis</h3>
|
||||
<div id="outline-container-org8bb1bfb" class="outline-3">
|
||||
<h3 id="org8bb1bfb"><span class="section-number-3">4.3.</span> H-Infinity Synthesis</h3>
|
||||
<div class="outline-text-3" id="text-4-3">
|
||||
<p>
|
||||
The generalized plant in Figure <a href="#org452ccad">14</a> containing the weighting functions is defined below.
|
||||
The generalized plant in Figure <a href="#orge2e8c55">14</a> containing the weighting functions is defined below.
|
||||
</p>
|
||||
|
||||
<div class="org-src-container">
|
||||
@ -995,7 +995,7 @@ And the standard \(\mathcal{H}_\infty\) synthesis using the <code>hinfsyn</code>
|
||||
</pre>
|
||||
</div>
|
||||
|
||||
<pre class="example" id="org162f0ab">
|
||||
<pre class="example" id="orgbeefca4">
|
||||
Resetting value of Gamma min based on D_11, D_12, D_21 terms
|
||||
|
||||
Test bounds: 0.1000 < gamma <= 1050.0000
|
||||
@ -1038,11 +1038,11 @@ H1 = 1 <span class="org-builtin">-</span> H2 <span class="org-builtin">-</span>
|
||||
</div>
|
||||
|
||||
<p>
|
||||
The bode plots of the three obtained complementary filters are shown in Figure <a href="#org4f0c4c6">16</a>.
|
||||
The bode plots of the three obtained complementary filters are shown in Figure <a href="#orgd7996d8">16</a>.
|
||||
</p>
|
||||
|
||||
|
||||
<div id="org4f0c4c6" class="figure">
|
||||
<div id="orgd7996d8" class="figure">
|
||||
<p><img src="figs/three_complementary_filters_results.png" alt="three_complementary_filters_results.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 16: </span>The three complementary filters obtained after \(\mathcal{H}_\infty\) synthesis</p>
|
||||
@ -1755,7 +1755,7 @@ H1 = 1 <span class="org-builtin">-</span> H2;
|
||||
</div>
|
||||
<div id="postamble" class="status">
|
||||
<p class="author">Author: Dehaeze Thomas</p>
|
||||
<p class="date">Created: 2021-09-01 mer. 16:59</p>
|
||||
<p class="date">Created: 2021-09-02 jeu. 10:04</p>
|
||||
</div>
|
||||
</body>
|
||||
</html>
|
||||
|
||||
@ -1,4 +1,4 @@
|
||||
#+TITLE: A new method of designing complementary filters for sensor fusion using the $\mathcal{H}_\infty$ synthesis - Matlab Computation
|
||||
#+TITLE: Designing Complementary filters for sensor fusion using $\mathcal{H}_\infty$ synthesis - Matlab Computation
|
||||
:DRAWER:
|
||||
#+LANGUAGE: en
|
||||
#+EMAIL: dehaeze.thomas@gmail.com
|
||||
@ -1442,4 +1442,7 @@ H1 = 1 - H2;
|
||||
#+end_src
|
||||
|
||||
* Bibliography :ignore:
|
||||
# For html export only:
|
||||
# bibliography:ref.bib
|
||||
|
||||
#+latex: \printbibliography
|
||||
|
||||
Binary file not shown.
@ -1,4 +1,4 @@
|
||||
% Created 2021-09-01 mer. 16:57
|
||||
% Created 2021-09-02 jeu. 10:03
|
||||
% Intended LaTeX compiler: pdflatex
|
||||
\documentclass[a4paper, 10pt, DIV=12, parskip=full]{scrreprt}
|
||||
\usepackage[utf8]{inputenc}
|
||||
@ -24,7 +24,7 @@
|
||||
\addbibresource{ref.bib}
|
||||
\author{Dehaeze Thomas}
|
||||
\date{\today}
|
||||
\title{A new method of designing complementary filters for sensor fusion using the \(\mathcal{H}_\infty\) synthesis - Matlab Computation}
|
||||
\title{Designing Complementary filters for sensor fusion using \(\mathcal{H}_\infty\) synthesis - Matlab Computation}
|
||||
\begin{document}
|
||||
|
||||
\maketitle
|
||||
@ -47,7 +47,6 @@ This document is divided into the following sections also corresponding to the p
|
||||
\chapter{H-Infinity synthesis of complementary filters}
|
||||
\label{sec:h_inf_synthesis_complementary_filters}
|
||||
\section{Synthesis Architecture}
|
||||
\label{sec:org7b3f474}
|
||||
In order to generate two complementary filters with a wanted shape, the generalized plant of Figure \ref{fig:h_infinity_robust_fusion_plant} can be used.
|
||||
The included weights \(W_1(s)\) and \(W_2(s)\) are used to specify the upper bounds of the complementary filters being generated.
|
||||
|
||||
@ -93,7 +92,6 @@ The presented synthesis method therefore allows to shape two filters \(H_1(s)\)
|
||||
The complete Matlab script for this part is given in Section \ref{sec:1_synthesis_complementary_filters}.
|
||||
|
||||
\section{Design of Weighting Function - Proposed formula}
|
||||
\label{sec:org69d0d73}
|
||||
A formula is proposed to help the design of the weighting functions:
|
||||
\begin{equation}
|
||||
W(s) = \left( \frac{
|
||||
@ -120,7 +118,6 @@ The general shape of a weighting function generated using the formula is shown i
|
||||
\end{figure}
|
||||
|
||||
\section{Weighting functions for the design of two complementary filters}
|
||||
\label{sec:org244c0a4}
|
||||
\label{sec:weighting_functions_example}
|
||||
|
||||
The weighting function formula \eqref{eq:weighting_function_formula} is used to generate the upper bounds of two complementary filters that we wish to design.
|
||||
@ -142,7 +139,6 @@ The inverse magnitude of these two weighting functions are shown in Figure \ref{
|
||||
\end{figure}
|
||||
|
||||
\section{Synthesis of the complementary filters}
|
||||
\label{sec:org3378b49}
|
||||
The generalized plant of Figure \ref{fig:h_infinity_robust_fusion_plant} is defined as follows:
|
||||
\begin{minted}[]{matlab}
|
||||
%% Generalized Plant
|
||||
@ -198,7 +194,6 @@ This function is described in Section \ref{sec:generateCF}.
|
||||
\end{minted}
|
||||
|
||||
\section{Obtained Complementary Filters}
|
||||
\label{sec:orgb52f55e}
|
||||
The obtained complementary filters are shown below and are found to be of order 5.
|
||||
Their bode plots are shown in figure \ref{fig:hinf_filters_results} and compare with the defined upper bounds.
|
||||
|
||||
@ -231,7 +226,6 @@ In this section, the proposed method for the design of complementary filters is
|
||||
The complete Matlab script for this part is given in Section \ref{sec:2_ligo_complementary_filters}.
|
||||
|
||||
\section{Specifications}
|
||||
\label{sec:orgc6bd7b2}
|
||||
The specifications for the set of complementary filters (\(L_1,H_1\)) used at the LIGO are summarized below (for further details, refer to \cite{hua04_polyp_fir_compl_filter_contr_system}):
|
||||
\begin{itemize}
|
||||
\item From 0 to 0.008 Hz, the magnitude \(|L_1(j\omega)|\) should be less or equal to \(8 \times 10^{-4}\)
|
||||
@ -249,7 +243,6 @@ The specifications are translated into upper bounds of the complementary filters
|
||||
\end{figure}
|
||||
|
||||
\section{FIR Filter}
|
||||
\label{sec:orgef72f87}
|
||||
To replicated the complementary filters developed in \cite{hua04_low_ligo}, the CVX Matlab toolbox \cite{grant14_cvx} is used.
|
||||
|
||||
The CVX toolbox is initialized and the \texttt{SeDuMi} solver \cite{sturm99_using_sedum} is used.
|
||||
@ -409,7 +402,6 @@ H = [exp(-j*kron(w'.*2*pi,[0:n-1]))]*h;
|
||||
\end{figure}
|
||||
|
||||
\section{Weighting function design}
|
||||
\label{sec:orgcbee7b7}
|
||||
The weightings function that will be used for the \(\mathcal{H}_\infty\) synthesis of the complementary filters are now designed.
|
||||
|
||||
These weights will determine the order of the obtained filters.
|
||||
@ -466,7 +458,6 @@ The inverse magnitude of the weighting functions are shown in Figure \ref{fig:li
|
||||
\end{figure}
|
||||
|
||||
\section{Synthesis of the complementary filters}
|
||||
\label{sec:org8f9014e}
|
||||
The generalized plant of figure \ref{fig:h_infinity_robust_fusion_plant} is defined.
|
||||
\begin{minted}[]{matlab}
|
||||
%% Generalized plant for the H-infinity Synthesis
|
||||
@ -533,7 +524,6 @@ The magnitude of the obtained filters as well as the requirements are shown in F
|
||||
\end{figure}
|
||||
|
||||
\section{Comparison of the FIR filters and synthesized filters}
|
||||
\label{sec:org6367ef0}
|
||||
|
||||
Let's now compare the FIR filters designed in \cite{hua04_low_ligo} with the with complementary filters obtained with the \(\mathcal{H}_\infty\) synthesis.
|
||||
|
||||
@ -558,7 +548,6 @@ In this section, the classical feedback architecture shown in Figure \ref{fig:fe
|
||||
The complete Matlab script for this part is given in Section \ref{sec:3_closed_loop_complementary_filters}.
|
||||
|
||||
\section{Weighting Function design}
|
||||
\label{sec:org856a399}
|
||||
Weighting functions using the \texttt{generateWF} Matlab function are designed to specify the upper bounds of the complementary filters to be designed.
|
||||
These weighting functions are the same as the ones used in Section \ref{sec:weighting_functions_example}.
|
||||
|
||||
@ -569,7 +558,6 @@ W2 = generateWF('n', 2, 'w0', 2*pi*10, 'G0', 1/10, 'Ginf', 1000, 'Gc', 0.45);
|
||||
\end{minted}
|
||||
|
||||
\section{Generalized plant}
|
||||
\label{sec:orgbba676f}
|
||||
The generalized plant of Figure \ref{fig:feedback_synthesis_architecture_generalized_plant} is defined below:
|
||||
\begin{minted}[]{matlab}
|
||||
%% Generalized plant for "closed-loop" complementary filter synthesis
|
||||
@ -584,7 +572,6 @@ P = [ W1 0 1;
|
||||
\end{figure}
|
||||
|
||||
\section{Synthesis of the closed-loop complementary filters}
|
||||
\label{sec:org44dc95f}
|
||||
And the standard \(\mathcal{H}_\infty\) synthesis is performed.
|
||||
\begin{minted}[]{matlab}
|
||||
%% Standard H-Infinity Synthesis
|
||||
@ -614,7 +601,6 @@ And the standard \(\mathcal{H}_\infty\) synthesis is performed.
|
||||
\end{verbatim}
|
||||
|
||||
\section{Synthesized filters}
|
||||
\label{sec:org9e70c62}
|
||||
The obtained filter \(L(s)\) can then be included in the feedback architecture shown in Figure \ref{fig:hinf_filters_results_mixed_sensitivity}.
|
||||
|
||||
The closed-loop transfer functions from \(\hat{x}_1\) to \(\hat{x}\) and from \(\hat{x}_2\) to \(\hat{x}\) corresponding respectively to the sensitivity and complementary sensitivity transfer functions are defined below:
|
||||
@ -651,7 +637,6 @@ In this section, the proposed synthesis method of complementary filters is gener
|
||||
The complete Matlab script for this part is given in Section \ref{sec:4_three_complementary_filters}.
|
||||
|
||||
\section{Synthesis Architecture}
|
||||
\label{sec:orgc4d55c9}
|
||||
The synthesis objective is to shape three filters that are complementary.
|
||||
This corresponds to the conditions \eqref{eq:obj_three_cf} where \(W_1(s)\), \(W_2(s)\) and \(W_3(s)\) are weighting functions used to specify the maximum wanted magnitude of the three complementary filters.
|
||||
|
||||
@ -680,7 +665,6 @@ The last filter \(H_1(s)\) is defined as the complementary of the two others as
|
||||
\end{equation}
|
||||
|
||||
\section{Weights}
|
||||
\label{sec:orgf866c06}
|
||||
The three weighting functions are defined as shown below.
|
||||
|
||||
\begin{minted}[]{matlab}
|
||||
@ -699,7 +683,6 @@ Their inverse magnitudes are displayed in Figure \ref{fig:three_weighting_functi
|
||||
\end{figure}
|
||||
|
||||
\section{H-Infinity Synthesis}
|
||||
\label{sec:org0db8a8c}
|
||||
|
||||
The generalized plant in Figure \ref{fig:comp_filter_three_hinf} containing the weighting functions is defined below.
|
||||
|
||||
|
||||
@ -3,9 +3,9 @@
|
||||
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
|
||||
<html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="en">
|
||||
<head>
|
||||
<!-- 2021-09-01 mer. 10:15 -->
|
||||
<!-- 2021-09-02 jeu. 10:03 -->
|
||||
<meta http-equiv="Content-Type" content="text/html;charset=utf-8" />
|
||||
<title>A new method of designing complementary filters for sensor fusion using the $\mathcal{H}_\infty$ synthesis - Tikz Figures</title>
|
||||
<title>Designing complementary filters for sensor fusion using $\mathcal{H}_\infty$ synthesis - Tikz Figures</title>
|
||||
<meta name="author" content="Thomas Dehaeze" />
|
||||
<meta name="generator" content="Org Mode" />
|
||||
<link rel="stylesheet" type="text/css" href="https://research.tdehaeze.xyz/css/style.css"/>
|
||||
@ -34,30 +34,30 @@
|
||||
|
|
||||
<a accesskey="H" href="../index.html"> HOME </a>
|
||||
</div><div id="content" class="content">
|
||||
<h1 class="title">A new method of designing complementary filters for sensor fusion using the \(\mathcal{H}_\infty\) synthesis - Tikz Figures</h1>
|
||||
<h1 class="title">Designing complementary filters for sensor fusion using \(\mathcal{H}_\infty\) synthesis - Tikz Figures</h1>
|
||||
<div id="table-of-contents" role="doc-toc">
|
||||
<h2>Table of Contents</h2>
|
||||
<div id="text-table-of-contents" role="doc-toc">
|
||||
<ul>
|
||||
<li><a href="#orgaddbc4b">1. Sensor Fusion - Overview</a></li>
|
||||
<li><a href="#orgcde34ed">2. Sensor Model</a></li>
|
||||
<li><a href="#org881c7cf">3. Sensor Model with calibration</a></li>
|
||||
<li><a href="#org52dba0e">4. Sensor Fusion Architecture</a></li>
|
||||
<li><a href="#org51c6b92">5. Sensor Model with Uncertainty</a></li>
|
||||
<li><a href="#orgd0c3320">6. Sensor Model with Uncertainty - Simplified</a></li>
|
||||
<li><a href="#orgd991dd2">7. Sensor fusion architecture with sensor dynamics uncertainty</a></li>
|
||||
<li><a href="#orgb45d1ac">8. Uncertainty set of the super sensor dynamics</a></li>
|
||||
<li><a href="#orgd194580">9. Generalized plant used for \(\mathcal{H}_\infty\) synthesis of complementary filters</a></li>
|
||||
<li><a href="#orgebe7ec5">10. Architecture used for \(\mathcal{H}_\infty\) synthesis of complementary filters</a></li>
|
||||
<li><a href="#orgf9a2f39">11. LIGO Sensor Fusion Architecture</a></li>
|
||||
<li><a href="#orga2611bb">12. Closed-Loop Complementary Filters</a></li>
|
||||
<li><a href="#org5af72e4">13. Closed-Loop Fusion Architecture</a></li>
|
||||
<li><a href="#org4e79ee7">14. Feedback Loop Sensor Fusion Architecture</a></li>
|
||||
<li><a href="#orgb59c88b">15. Feedback Sensor Fusion - Generalized Plant</a></li>
|
||||
<li><a href="#orge593f6d">16. Sensor Fusion - Sequential</a></li>
|
||||
<li><a href="#org3b323e8">17. Sensor Fusion - Parallel</a></li>
|
||||
<li><a href="#org9baa5b5">18. Architecture for \(\mathcal{H}_\infty\) synthesis of three complementary filters</a></li>
|
||||
<li><a href="#org10ffb09">19. Architecture for \(\mathcal{H}_\infty\) synthesis of three complementary filters</a></li>
|
||||
<li><a href="#org50515cc">1. Sensor Fusion - Overview</a></li>
|
||||
<li><a href="#orgd0f42e8">2. Sensor Model</a></li>
|
||||
<li><a href="#orgfe2b210">3. Sensor Model with calibration</a></li>
|
||||
<li><a href="#org7a71f40">4. Sensor Fusion Architecture</a></li>
|
||||
<li><a href="#org02ee7f2">5. Sensor Model with Uncertainty</a></li>
|
||||
<li><a href="#org15f5c37">6. Sensor Model with Uncertainty - Simplified</a></li>
|
||||
<li><a href="#org6703507">7. Sensor fusion architecture with sensor dynamics uncertainty</a></li>
|
||||
<li><a href="#org730783f">8. Uncertainty set of the super sensor dynamics</a></li>
|
||||
<li><a href="#org38b031f">9. Generalized plant used for \(\mathcal{H}_\infty\) synthesis of complementary filters</a></li>
|
||||
<li><a href="#orgc8963ca">10. Architecture used for \(\mathcal{H}_\infty\) synthesis of complementary filters</a></li>
|
||||
<li><a href="#orgc9e8b5c">11. LIGO Sensor Fusion Architecture</a></li>
|
||||
<li><a href="#org852a702">12. Closed-Loop Complementary Filters</a></li>
|
||||
<li><a href="#org21e2e13">13. Closed-Loop Fusion Architecture</a></li>
|
||||
<li><a href="#org784c1ac">14. Feedback Loop Sensor Fusion Architecture</a></li>
|
||||
<li><a href="#org08b29ce">15. Feedback Sensor Fusion - Generalized Plant</a></li>
|
||||
<li><a href="#org9bf391e">16. Sensor Fusion - Sequential</a></li>
|
||||
<li><a href="#orgc2ef6aa">17. Sensor Fusion - Parallel</a></li>
|
||||
<li><a href="#org90ea882">18. Architecture for \(\mathcal{H}_\infty\) synthesis of three complementary filters</a></li>
|
||||
<li><a href="#orge8d8968">19. Architecture for \(\mathcal{H}_\infty\) synthesis of three complementary filters</a></li>
|
||||
</ul>
|
||||
</div>
|
||||
</div>
|
||||
@ -65,8 +65,8 @@
|
||||
Configuration file is accessible <a href="config.html">here</a>.
|
||||
</p>
|
||||
|
||||
<div id="outline-container-orgaddbc4b" class="outline-2">
|
||||
<h2 id="orgaddbc4b"><span class="section-number-2">1.</span> Sensor Fusion - Overview</h2>
|
||||
<div id="outline-container-org50515cc" class="outline-2">
|
||||
<h2 id="org50515cc"><span class="section-number-2">1.</span> Sensor Fusion - Overview</h2>
|
||||
<div class="outline-text-2" id="text-1">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate">\definecolor</span>{myblue}{rgb}{0, 0.447, 0.741}
|
||||
@ -106,7 +106,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="orgfdc79fa" class="figure">
|
||||
<div id="org3c4bfaf" class="figure">
|
||||
<p><img src="figs/sensor_fusion_overview.png" alt="sensor_fusion_overview.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 1: </span>Schematic of a sensor fusion architecture using complementary filters</p>
|
||||
@ -114,8 +114,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orgcde34ed" class="outline-2">
|
||||
<h2 id="orgcde34ed"><span class="section-number-2">2.</span> Sensor Model</h2>
|
||||
<div id="outline-container-orgd0f42e8" class="outline-2">
|
||||
<h2 id="orgd0f42e8"><span class="section-number-2">2.</span> Sensor Model</h2>
|
||||
<div class="outline-text-2" id="text-2">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -136,7 +136,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="orga6c4cb8" class="figure">
|
||||
<div id="org9ced2bb" class="figure">
|
||||
<p><img src="figs/sensor_model.png" alt="sensor_model.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 2: </span>Basic sensor model consisting of a noise input \(n_i\) and a linear time invariant transfer function \(G_i(s)\)</p>
|
||||
@ -144,8 +144,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org881c7cf" class="outline-2">
|
||||
<h2 id="org881c7cf"><span class="section-number-2">3.</span> Sensor Model with calibration</h2>
|
||||
<div id="outline-container-orgfe2b210" class="outline-2">
|
||||
<h2 id="orgfe2b210"><span class="section-number-2">3.</span> Sensor Model with calibration</h2>
|
||||
<div class="outline-text-2" id="text-3">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -171,7 +171,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="orgd38eff7" class="figure">
|
||||
<div id="org2edc54e" class="figure">
|
||||
<p><img src="figs/sensor_model_calibrated.png" alt="sensor_model_calibrated.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 3: </span>Normalized sensors using the inverse of an estimate \(\hat{G}_i(s)\) of the sensor dynamics</p>
|
||||
@ -179,8 +179,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org52dba0e" class="outline-2">
|
||||
<h2 id="org52dba0e"><span class="section-number-2">4.</span> Sensor Fusion Architecture</h2>
|
||||
<div id="outline-container-org7a71f40" class="outline-2">
|
||||
<h2 id="org7a71f40"><span class="section-number-2">4.</span> Sensor Fusion Architecture</h2>
|
||||
<div class="outline-text-2" id="text-4">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate">\definecolor</span>{myblue}{rgb}{0, 0.447, 0.741}
|
||||
@ -232,7 +232,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org63acf85" class="figure">
|
||||
<div id="org895e3d1" class="figure">
|
||||
<p><img src="figs/fusion_super_sensor.png" alt="fusion_super_sensor.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 4: </span>Sensor fusion architecture with two normalized sensors</p>
|
||||
@ -240,8 +240,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org51c6b92" class="outline-2">
|
||||
<h2 id="org51c6b92"><span class="section-number-2">5.</span> Sensor Model with Uncertainty</h2>
|
||||
<div id="outline-container-org02ee7f2" class="outline-2">
|
||||
<h2 id="org02ee7f2"><span class="section-number-2">5.</span> Sensor Model with Uncertainty</h2>
|
||||
<div class="outline-text-2" id="text-5">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -275,7 +275,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org937c435" class="figure">
|
||||
<div id="orga300be8" class="figure">
|
||||
<p><img src="figs/sensor_model_uncertainty.png" alt="sensor_model_uncertainty.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 5: </span>Sensor with multiplicative input uncertainty</p>
|
||||
@ -283,8 +283,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orgd0c3320" class="outline-2">
|
||||
<h2 id="orgd0c3320"><span class="section-number-2">6.</span> Sensor Model with Uncertainty - Simplified</h2>
|
||||
<div id="outline-container-org15f5c37" class="outline-2">
|
||||
<h2 id="org15f5c37"><span class="section-number-2">6.</span> Sensor Model with Uncertainty - Simplified</h2>
|
||||
<div class="outline-text-2" id="text-6">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -311,7 +311,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="orgce8343e" class="figure">
|
||||
<div id="org7ccc258" class="figure">
|
||||
<p><img src="figs/sensor_model_uncertainty_simplified.png" alt="sensor_model_uncertainty_simplified.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 6: </span>Simplified sensor model</p>
|
||||
@ -319,8 +319,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orgd991dd2" class="outline-2">
|
||||
<h2 id="orgd991dd2"><span class="section-number-2">7.</span> Sensor fusion architecture with sensor dynamics uncertainty</h2>
|
||||
<div id="outline-container-org6703507" class="outline-2">
|
||||
<h2 id="org6703507"><span class="section-number-2">7.</span> Sensor fusion architecture with sensor dynamics uncertainty</h2>
|
||||
<div class="outline-text-2" id="text-7">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate">\definecolor</span>{myblue}{rgb}{0, 0.447, 0.741}
|
||||
@ -380,7 +380,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org5b69e5a" class="figure">
|
||||
<div id="orgc80e315" class="figure">
|
||||
<p><img src="figs/sensor_fusion_dynamic_uncertainty.png" alt="sensor_fusion_dynamic_uncertainty.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 7: </span>Sensor fusion architecture with sensor dynamics uncertainty</p>
|
||||
@ -388,8 +388,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orgb45d1ac" class="outline-2">
|
||||
<h2 id="orgb45d1ac"><span class="section-number-2">8.</span> Uncertainty set of the super sensor dynamics</h2>
|
||||
<div id="outline-container-org730783f" class="outline-2">
|
||||
<h2 id="org730783f"><span class="section-number-2">8.</span> Uncertainty set of the super sensor dynamics</h2>
|
||||
<div class="outline-text-2" id="text-8">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate">\definecolor</span>{myblue}{rgb}{0, 0.447, 0.741}
|
||||
@ -432,7 +432,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="orgc80ec14" class="figure">
|
||||
<div id="org6687171" class="figure">
|
||||
<p><img src="figs/uncertainty_set_super_sensor.png" alt="uncertainty_set_super_sensor.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 8: </span>Uncertainty region of the super sensor dynamics in the complex plane (grey circle). The contribution of both sensors 1 and 2 to the total uncertainty are represented respectively by a blue circle and a red circle. The frequency dependency \(\omega\) is here omitted.</p>
|
||||
@ -440,8 +440,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orgd194580" class="outline-2">
|
||||
<h2 id="orgd194580"><span class="section-number-2">9.</span> Generalized plant used for \(\mathcal{H}_\infty\) synthesis of complementary filters</h2>
|
||||
<div id="outline-container-org38b031f" class="outline-2">
|
||||
<h2 id="org38b031f"><span class="section-number-2">9.</span> Generalized plant used for \(\mathcal{H}_\infty\) synthesis of complementary filters</h2>
|
||||
<div class="outline-text-2" id="text-9">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -473,7 +473,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org725e946" class="figure">
|
||||
<div id="orge10d26f" class="figure">
|
||||
<p><img src="figs/h_infinity_robust_fusion_plant.png" alt="h_infinity_robust_fusion_plant.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 9: </span>Generalized plant used for \(\mathcal{H}_\infty\) synthesis of complementary filters</p>
|
||||
@ -481,8 +481,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orgebe7ec5" class="outline-2">
|
||||
<h2 id="orgebe7ec5"><span class="section-number-2">10.</span> Architecture used for \(\mathcal{H}_\infty\) synthesis of complementary filters</h2>
|
||||
<div id="outline-container-orgc8963ca" class="outline-2">
|
||||
<h2 id="orgc8963ca"><span class="section-number-2">10.</span> Architecture used for \(\mathcal{H}_\infty\) synthesis of complementary filters</h2>
|
||||
<div class="outline-text-2" id="text-10">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -514,7 +514,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org47d276f" class="figure">
|
||||
<div id="org0646c95" class="figure">
|
||||
<p><img src="figs/h_infinity_robust_fusion_fb.png" alt="h_infinity_robust_fusion_fb.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 10: </span>Generalized plant with the synthesized filter</p>
|
||||
@ -522,8 +522,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orgf9a2f39" class="outline-2">
|
||||
<h2 id="orgf9a2f39"><span class="section-number-2">11.</span> LIGO Sensor Fusion Architecture</h2>
|
||||
<div id="outline-container-orgc9e8b5c" class="outline-2">
|
||||
<h2 id="orgc9e8b5c"><span class="section-number-2">11.</span> LIGO Sensor Fusion Architecture</h2>
|
||||
<div class="outline-text-2" id="text-11">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate">\definecolor</span>{myblue}{rgb}{0, 0.447, 0.741}
|
||||
@ -570,7 +570,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org81fe1cf" class="figure">
|
||||
<div id="org845a82e" class="figure">
|
||||
<p><img src="figs/ligo_super_sensor_architecture.png" alt="ligo_super_sensor_architecture.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 11: </span>Simplified block diagram of the sensor blending strategy for the first stage at the LIGO</p>
|
||||
@ -578,8 +578,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orga2611bb" class="outline-2">
|
||||
<h2 id="orga2611bb"><span class="section-number-2">12.</span> Closed-Loop Complementary Filters</h2>
|
||||
<div id="outline-container-org852a702" class="outline-2">
|
||||
<h2 id="org852a702"><span class="section-number-2">12.</span> Closed-Loop Complementary Filters</h2>
|
||||
<div class="outline-text-2" id="text-12">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -602,7 +602,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org57ca8ce" class="figure">
|
||||
<div id="org89c3b66" class="figure">
|
||||
<p><img src="figs/feedback_sensor_fusion.png" alt="feedback_sensor_fusion.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 12: </span>“Closed-Loop” complementary filters</p>
|
||||
@ -610,8 +610,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org5af72e4" class="outline-2">
|
||||
<h2 id="org5af72e4"><span class="section-number-2">13.</span> Closed-Loop Fusion Architecture</h2>
|
||||
<div id="outline-container-org21e2e13" class="outline-2">
|
||||
<h2 id="org21e2e13"><span class="section-number-2">13.</span> Closed-Loop Fusion Architecture</h2>
|
||||
<div class="outline-text-2" id="text-13">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -644,7 +644,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org8886809" class="figure">
|
||||
<div id="org7c31417" class="figure">
|
||||
<p><img src="figs/feedback_sensor_fusion_arch.png" alt="feedback_sensor_fusion_arch.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 13: </span>Classical feedback architecture used for sensor fusion</p>
|
||||
@ -652,8 +652,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org4e79ee7" class="outline-2">
|
||||
<h2 id="org4e79ee7"><span class="section-number-2">14.</span> Feedback Loop Sensor Fusion Architecture</h2>
|
||||
<div id="outline-container-org784c1ac" class="outline-2">
|
||||
<h2 id="org784c1ac"><span class="section-number-2">14.</span> Feedback Loop Sensor Fusion Architecture</h2>
|
||||
<div class="outline-text-2" id="text-14">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -675,7 +675,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org705140c" class="figure">
|
||||
<div id="org3aa7290" class="figure">
|
||||
<p><img src="figs/feedback_synthesis_architecture.png" alt="feedback_synthesis_architecture.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 14: </span>Feedback architecture with included weights</p>
|
||||
@ -683,8 +683,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orgb59c88b" class="outline-2">
|
||||
<h2 id="orgb59c88b"><span class="section-number-2">15.</span> Feedback Sensor Fusion - Generalized Plant</h2>
|
||||
<div id="outline-container-org08b29ce" class="outline-2">
|
||||
<h2 id="org08b29ce"><span class="section-number-2">15.</span> Feedback Sensor Fusion - Generalized Plant</h2>
|
||||
<div class="outline-text-2" id="text-15">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -716,7 +716,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org32b295e" class="figure">
|
||||
<div id="orgc836b02" class="figure">
|
||||
<p><img src="figs/feedback_synthesis_architecture_generalized_plant.png" alt="feedback_synthesis_architecture_generalized_plant.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 15: </span>Generalized plant used for the \(\mathcal{H}_\infty\) synthesis of “closed-loop” complementary filters</p>
|
||||
@ -724,8 +724,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-orge593f6d" class="outline-2">
|
||||
<h2 id="orge593f6d"><span class="section-number-2">16.</span> Sensor Fusion - Sequential</h2>
|
||||
<div id="outline-container-org9bf391e" class="outline-2">
|
||||
<h2 id="org9bf391e"><span class="section-number-2">16.</span> Sensor Fusion - Sequential</h2>
|
||||
<div class="outline-text-2" id="text-16">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -767,7 +767,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org6fe58a4" class="figure">
|
||||
<div id="org187b8ba" class="figure">
|
||||
<p><img src="figs/sensor_fusion_three_sequential.png" alt="sensor_fusion_three_sequential.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 16: </span>Sequential fusion</p>
|
||||
@ -775,8 +775,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org3b323e8" class="outline-2">
|
||||
<h2 id="org3b323e8"><span class="section-number-2">17.</span> Sensor Fusion - Parallel</h2>
|
||||
<div id="outline-container-orgc2ef6aa" class="outline-2">
|
||||
<h2 id="orgc2ef6aa"><span class="section-number-2">17.</span> Sensor Fusion - Parallel</h2>
|
||||
<div class="outline-text-2" id="text-17">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"><span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -813,7 +813,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org6ec849a" class="figure">
|
||||
<div id="org72ff7bd" class="figure">
|
||||
<p><img src="figs/sensor_fusion_three_parallel.png" alt="sensor_fusion_three_parallel.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 17: </span>Parallel fusion</p>
|
||||
@ -821,8 +821,8 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="outline-container-org9baa5b5" class="outline-2">
|
||||
<h2 id="org9baa5b5"><span class="section-number-2">18.</span> Architecture for \(\mathcal{H}_\infty\) synthesis of three complementary filters</h2>
|
||||
<div id="outline-container-org90ea882" class="outline-2">
|
||||
<h2 id="org90ea882"><span class="section-number-2">18.</span> Architecture for \(\mathcal{H}_\infty\) synthesis of three complementary filters</h2>
|
||||
<div class="outline-text-2" id="text-18">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"> <span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -865,15 +865,15 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org7b3470a" class="figure">
|
||||
<div id="org7b7c9c7" class="figure">
|
||||
<p><img src="figs/comp_filter_three_hinf_gen_plant.png" alt="comp_filter_three_hinf_gen_plant.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 18: </span>Generalized plant for the \(\mathcal{H}_\infty\) synthesis of three complementary filters</p>
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
<div id="outline-container-org10ffb09" class="outline-2">
|
||||
<h2 id="org10ffb09"><span class="section-number-2">19.</span> Architecture for \(\mathcal{H}_\infty\) synthesis of three complementary filters</h2>
|
||||
<div id="outline-container-orge8d8968" class="outline-2">
|
||||
<h2 id="orge8d8968"><span class="section-number-2">19.</span> Architecture for \(\mathcal{H}_\infty\) synthesis of three complementary filters</h2>
|
||||
<div class="outline-text-2" id="text-19">
|
||||
<div class="org-src-container">
|
||||
<pre class="src src-latex"> <span class="org-font-latex-sedate"><span class="org-keyword">\begin</span></span>{<span class="org-function-name">tikzpicture</span>}
|
||||
@ -916,7 +916,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
|
||||
|
||||
<div id="org71495ac" class="figure">
|
||||
<div id="org4f6f8fa" class="figure">
|
||||
<p><img src="figs/comp_filter_three_hinf_fb.png" alt="comp_filter_three_hinf_fb.png" />
|
||||
</p>
|
||||
<p><span class="figure-number">Figure 19: </span>Generalized plant with the synthesized filter for the \(\mathcal{H}_\infty\) synthesis of three complementary filters</p>
|
||||
@ -926,7 +926,7 @@ Configuration file is accessible <a href="config.html">here</a>.
|
||||
</div>
|
||||
<div id="postamble" class="status">
|
||||
<p class="author">Author: Thomas Dehaeze</p>
|
||||
<p class="date">Created: 2021-09-01 mer. 10:15</p>
|
||||
<p class="date">Created: 2021-09-02 jeu. 10:03</p>
|
||||
</div>
|
||||
</body>
|
||||
</html>
|
||||
|
||||
@ -1,4 +1,4 @@
|
||||
#+TITLE: A new method of designing complementary filters for sensor fusion using the $\mathcal{H}_\infty$ synthesis - Tikz Figures
|
||||
#+TITLE: Designing complementary filters for sensor fusion using $\mathcal{H}_\infty$ synthesis - Tikz Figures
|
||||
:DRAWER:
|
||||
#+HTML_LINK_HOME: ../index.html
|
||||
#+HTML_LINK_UP: ../index.html
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user