Supervisory authorities

CNRS Ecole Centrale de Lyon Université de Lyon Université Lyon 1 INSA de Lyon

Our partners



Home > Events > Archive des séminaires Ampère

Séminaires Ampère 2018

by Laurent Krähenbühl - published on , updated on

Liens vers les séminaires de 2021, 2020, 2019, 2018, 2017, 2016, 2015, 2014, 2013, 2012, 2011, 2010

05/12/2018, 12h30-14h00 : Federico Morelli (Ampère) - PhD Club

ECL, H9, salle de Conférence

Optimal Experiment Design for the Identification of One Module in the Interconnection of Locally controlled Systems

In this presentation, we consider the problem of designing the least costly experiment that leads to a sufficiently accurate estimate of one module in a network of locally controlled systems. A module in such a network can be identified by exciting the corresponding local closed loop system. Such an excitation signal will not only perturb the input/output of the to-be-identified module, but also other modules due to the interconnection. Consequently, the cost of the identification can be expressed as the sum of the influence of the excitation signal on the inputs and outputs of all locally controlled systems. We develop a methodology to design the spectrum of the excitation signal in such a way that this cost is minimized while guaranteeing a certain accuracy for the identified model. We also propose an alternative identification configuration which can further reduce the propagation of the excitation signal to other modules and we make steps to robustify this optimal experiment design problem with respect to the cost of the identification.

28/11/2018, 14h00-15h30 : Zohra Kader (post-doc L2S)

ECL, H9, salle de Conférence

Control of switched systems

In this talk, we first consider the problem of stabilization of switched systems with state-dependent switching laws. A general result is proposed for the case of input-affine nonlinear systems. A full state switching controller is designed in order to ensure the local asymptotic stability of the closed-loop system. Then, the result is applied to the particular case of switched affine systems. A constructive method based on LMI conditions is given in order to design nonlinear switching surfaces and provide an estimation of a non-ellipsoidal domain of attraction. In addition, the approach is extended to robust switching laws design when the measurements are affected by a bounded disturbance. Linear switching surfaces depending on the perturbed measurements are designed. Secondly, we consider the problem of symbolic model design for the class of incrementally stable switched systems with time-dependent switching laws. Contrarily to the existing results in the literature where switching is considered as periodically controlled, here, we consider aperiodic time sampling resulting either from uncertain or event-based sampling mechanisms. Finally, the talk will end with a summary of different proposed approaches and a discussion about open problems.

13/11/2018, 12h30-14-00 : Sigurd Jakobsen, NTNU (Trondheim, Norvège)

ECL, H9, salle de Conférence

Identification of hydro power plant dynamics using PMU measurements

In recent years there has been a growing concern that the frequency quality in the Nordic power system is deteriorating. Due to this new requirements for the connection of power plants have been developed. In these requirements there are described extensive tests for validating the models and performance of the plants. As a response to the new requirements we at NTNU have looked into if one can do some of the tests using measurements from the power system during normal operation instead of the extensive offline tests that are required. In this talk I will present the work we did on this together with Xavier here in Lyon.

06/11/2018, 12h30-14h00 : Georgios Birpoutsoukis (Séminaire MIS)

ECL, H9, salle TIC

Nonparametric model estimation for dynamic systems and structure detection

Across different fields of science, mathematical models of physical and technical processes have been becoming increasingly necessary and they are typically used in applications such as process simulation, prediction and control. Despite of the fact that different applications are accompanied by different model requirements and performance criteria, the choice of the model structure lies in the core of the modeling problem and constitutes a crucial step of the modeling procedure.
One way to infer information about the structure of the system under study is by the use of the so-called nonparametric models. The latter are characterized by a flexible model structure and they can be seen as an X-ray of the system since they can potentially provide information about the structure of the system under study. However, they usually consist of a large number of parameters, a fact that constitutes the model estimation quite difficult, especially when the complexity of the modeled dynamics increases. Moreover, the path from a general flexible nonparametric model to a more customized-to-the-application parametric one remains an open problem in many cases.
In this presentation, I am giving an overview of my project proposal in terms of my application for a CNRS researcher position. The proposed project consists of basically two research directions. The first one includes the improvement of the latest system identification methods which employ tools from the machine learning community in order to tackle the dimensionality issues in the nonparametric model case. The second direction deals with structure detection once a nonparametric model has been estimated. For both directions, we will go through the latest research trends, we will see what has been achieved until now and we will identify the missing pieces of the puzzle.

23/10/2018, 12h30-14h00 : Fabricio Saggin (PhD Club)

ECL, H9, salle de Conférence

Robust Control for MEMS Inertial Sensors

Micro-electro-mechanical systems (MEMS) inertial sensors are micromachined devices that have been widely employed to measure linear motion and rotation in numerous applications. This kind of sensors has the advantage to be easily integrated into electronic devices, cheap and low power consuming. However, compared to traditional mechanical or optical inertial sensors, MEMS sensors have a degraded precision.
The working principle of MEMS sensors relies on controlling and detecting the oscillation characteristics of an undamped resonant element. Thus, the better the oscillation control, the better the sensor performance. In this context, we propose two different control strategies: the direct control, based on the direct measurements of the oscillations (a modulated signal); and the envelope-based approach, that relies on the measurements of amplitude and phase of the oscillations (low-frequency signals). We discuss the advantages, drawbacks, and links between these approaches. Moreover, we present recent results of practical implementation that validate the direct control approach.

27/09/2018, 14h00-15-30 : Srinivas Tadigadapa, (Northeastern University, Boston)

Polytech (Doua), amphithéâtre

Microresonators for biosensing applications
The field of micro and nano sensors has rapidly evolved over the last couple of decades with many of the highest performance devices exploiting the phenomenon of resonance for sensing. Micro and nano scale designs, fabrication, and integration of new materials offer unique opportunities for innovative, novel, and robust sensor configurations. Examples of micromachined electromechanical, acoustic, optical, and nanoscale resonator devices will be presented. This talk will specifically focus on resonator based chemical and biological sensors will be discussed. An overview of critical design considerations such as resonator geometry, the Q-factor, and performance advantages of these devices will be presented.

20/07/2018, 10h00-11-30 : Rifat Sipahi, (Northeastern University, Boston)

INSA, Saint-Exupéry, 1er étage, salle de réunion Ampère

Effects of Delays on Human-in-the-Loop Dynamics and Network Systems
In this talk, we will present our recent results on how delays can influence human-in-the-loop dynamics and network systems, primarily from stability point of view. We will first study the problem of a human operator performing a tracking task with a machine, where human reaction delay affects the tracking performance.
An adaptive controller design will next be proposed in order to provide assistance to the human, limitations in the controller design will be discussed. Next, we will study a consensus dynamics in which time delay influences agents’ decisions and hence the group dynamics. In this dynamics, consensus can be achieved for only up to a certain amount of delay, called the "delay margin". By removing certain edges from the underlying graph of the consensus dynamics, it will become possible that delay margin, thus the amount of delay that can be tolerated, can be increased.
A systematic approach will be presented regarding how many edges and which edges to remove such that delay margin can be improved.

27/06/2018 : Journée annuelle du Département MIS.

Lieu : Château de Saint-Bernard.

Matin : Huit doctorants en seconde année de thèse présentent leurs travaux.
Après-midi : réflexion prospective sur l’avenir des thématiques scientifiques du Département.

26/06/2018, 12h30-14h00 : Kévin Colin (PhD Club)

ECL, H9, salle de Conférence

System identification of MEMS sensors : application to the gyroscope

In mechanical control of moving objects one has to decide of the sensors to use. This choice is always based on the trade-off efficiency of the measurement/cost. Some advanced techniques as optical sensors are used but are very cumbersome and expensive. A less cumbersome, less expensive but also less accurate way to measure a motion is to use MEMS (Microelectromechanical systems) technologies. There is an active research on this type of sensors to improve their efficiency of measurement by using automatic control techniques. This is the aim of the Ampere lab project NEXT4MEMS. To develop these control techniques we need an accurate model of the sensors. This presentation will focus on the MEMS gyroscope modeling.

In the literature one can find a majority of papers using grey-box models, based on mechanical equations and information from experimental data to compute unknown parameters. The angular rate of a moving system is then deduced by using these models and the control techniques. However these papers do not take account of the parasite effect of the implementation of a MEMS gyroscope on an electronic card which can alterate the deduction. The aim of this PhD club is to present how black-box Prediction-Error-Modeling (PEM) can be used to improve the model of MEMS gyroscopes. However the parameters of the computed models have always uncertainties due to noise and they can affect the angular rate deduction. One field of System Identification, Experiment Design, can be used to reduce or at least to redistribute the uncertainties on the parameters such that we minimize the angular rate deduction error.

18/06/2018, 12h00-14-00 : Prof. Fabian Wirth, Univ. de Passau (séminaire Control@Lyon)

INSA, Saint-Exupéry, M1B

Stabilization of differential-algebraic systems by switching
One possible way of stabilizing switched linear systems is to choose the switching instances in such a manner that the resulting time-varying system is exponentially stable. It is a classical result for switched linear ordinary differential equations that this is possible if there exists a Hurwitz matrix in the convex hull of the constituent matrices of the switched linear system.

For switched differential algebraic equations this result cannot be extended directly. On the one hand the switching may induce impulsive components of solutions which destroy stability. On the other hand it is by no means clear how the concept of convex hull can be transferred to the differential algebraic case. In particular with the aim of obtaining a meaningful construction from the point of view of dynamical systems.

In the talk we will discuss several sufficient conditions which allow the construction of stabilizing switching sequences for DAEs. One approach relies on the approximation of the DAE dynamics by systems of ordinary differential equations with fast dynamics, i.e. singularly perturbed systems. Another method uses estimates for the discontinuities induced by switching to obtain stabilizing switching sequences.

The talk is based on joint work with Andrii Mironchenko and Kai Wulff.

14/06/2018 : Journée annuelle du Département EE.

Lieu : "Musée de l’Aviation Clément Ader" à Corbas

Matin : présentation des doctorants deuxième année

  • BEYE Mamadou Lamine : Etude d’un bras d’onduleur Haute Tension à base de technologie GaN embarquant sa commande.
  • DAVID Romain : Study and design of a high efficiency laser diode with a fast current-mode control in an advanced CMOS technology for indirect time-of-flight applications
  • DEMUMIEUX Pierre : Etude et développement d’un convertisseur DC/DC 7.5kW 1MHz en composants GaN pour des applications dans l’aérospatiale
  • FOUINEAU Alexis : Méthodologies de conception de Transformateurs Moyennes Fréquences pour application aux réseaux haute tension et réseaux ferroviaires

Après-midi : visite du musée de l’air.

18/05/2018, 10h15 : Djidula MOTCHON (CRIStAL et LPP)

Lieu : ECL, H9, salle TIC

Stabilité des systèmes non-linéaires à commutation avec retards de détection des modes actifs

Les systèmes à commutation constituent une classe importante des systèmes dynamiques hybrides. Ils sont décrits par un ensemble de sous-systèmes ou modes et un signal de commutation qui indique à chaque instant le mode actif. Des techniques de commande dépendant du signal de commutation (mode-dependent controllers) ont été développées dans la littérature ces dernières années pour ces systèmes. Dans la pratique, le signal de commutation n’est pas connu, et pour ces méthodes de commande, une estimation du signal de commutation qui est généralement obtenue avec des retards de détection des modes actifs est utilisée. Ces retards de détection ont une influence sur la stabilité des systèmes à commutation car la ”vraie” loi de commande n’est pas appliquée en temps réel. Dans cet exposé, je vous présenterai les résultats de stabilité que nous avons établis récemment pour les systèmes non-linéaires à commutation avec retards de détection des modes actifs. Ils généralisent les conditions de stabilité que nous avons aussi établies pour la classe des systèmes linéaires à commutation.

18/05/2018, 14h00 : Ibtissem BOURAOUI

Lieu : ECL, H9, salle TIC

Synthèse d’observateurs pour les systèmes non linéaires

Dans cet exposé, je présenterai la synthèse d’observateurs d’état et adaptatifs pour des classes de systèmes non linéaires avec des mesures échantillonnées. Je proposerai un observateur continu-discret pour une classe de systèmes comportant des incertitudes et avec des sorties échantillonnées. Cet observateur sera dans un premier temps présenté sous forme impulsionnel avant d’être mis sous forme prédictive de sortie. Ensuite, je proposerai un observateur adaptatif pour une classe de systèmes comportant des coefficients constants inconnus linéairement paramétrés pour l’estimation de ces coefficients simultanément avec l’état. La synthèse de l’observateur initialement conçu pour fonctionner avec des sorties continues sera ensuite étendue au cas de la paramétrisation non linéaire et des sorties échantillonnées.

17/05/2018, 10h15 : Rémi AZOUIT (Sherbrooke Univ.)

Lieu : ECL, H9, salle TIC

Model reduction for open quantum systems

The dynamics of quantum systems interacting with its environment is governed by the Lindblad master equation. When the system is composed of several connected subsystem, following the evolution of this linear ordinary differential equation (ODE) is tedious due to the very large number of variables. Also for gaining physical insight of particular phenomenon within the whole system, it is necessary to develop rigorous methods to compute a reduced model based on some assumptions. In typical cases, the time scale in a subsystem of interest is much slower than in other subsystems (e.g a fast system acting as a "controller" coupled to a slower one). We will present a time-scale reduction technique consisting in eliminating the fast converging dynamics in order to focus on the slow dynamics. While this elimination is standard through perturbation theory in classical systems described by linear ODE, the need to ensure a reduced model conveying a physical interpretation is crucial here. Therefore, we propose a method based on geometric singular perturbation theory to compute the reduced model. This method ensure the preservation of the quantum structure by first, ensuring a Lindblad-form master equation for the reduced model and second that the parametrisation of the slow manifold is expressed as a completely positive map. We illustrate our method on several examples of physical

17/05/2018, 12h30 : Alessio IOVINE (EFFICACITY)

Lieu : ECL, H9, salle TIC

Techniques de contrôle pour les systèmes complexes: les cas des réseaux électriques intelligents et des véhicules autonomes

L’utilisation de techniques de contrôle complexes prend aujourd’hui de plus en plus d’importance dans notre vie quotidienne, car nous avons besoin de systèmes toujours plus flexibles et fiables, capables de faire face aux incertitudes et aux intermittences provenant d’un monde de plus en plus complexe et connecté.
En utilisant des contrôles non linéaires ou hybrides, le séminaire montrera comment interconnecter, contrôler et piloter des systèmes composés d’un certain nombre d’éléments dans les domaines des réseaux électriques intelligents et du contrôle du trafic. En particulier, il traitera de micro-réseaux en courant continu et véhicules autonomes.

17/05/2018, 16h15 : Anh-Tu Nguyen (LAMIH)

Lieu : ECL, H9, salle TIC

Constrained Nonlinear Control and Applications

This talk presents a brief overview of my research activities. The focus is put on the Takagi-Sugeno (T-S) model-based control approach and its applications. To this end, I will first provide a general idea on how the stability analysis and control design of nonlinear systems can be reformulated as convex optimization problems via T-S control framework. Then, the control design of nonlinear systems subject to various physical limitations on both system state and control input are discussed in more details. The advantages and drawbacks of this nonlinear control approach will be also highlighted. Finally, the effectiveness of T-S model-based control is illustrated with some real-world applications.

16/05/2018, 14h00 : Zohra KADER (L2S)

Lieu : ECL, H9, salle TIC

Control and observation of switched affine systems

In this talk we consider the problem of non-quadratic stabilization of switched systems. First, a general result is proposed for the case of nonlinear systems. A full state switching controller is designed in order to ensure the local asymptotic stability of the closed-loop system. Then, the result is applied to the particular case of switched affine systems. A constructive method based on LMI conditions is given in order to design nonlinear switching surfaces and provide an estimation of a non-ellipsoidal domain of attraction. In addition, the approach is extended to observer-based switching laws design. Both linear and nonlinear switching surfaces dependent on the estimated state are designed while using a Luenberger observer. Finally, illustrative examples are proposed in order to show the efficiency of the proposed methods and simulations are performed for a Buck converter structure.

02/05/2018 : Giacomo Casadei (GIPSA-Lab)

Lieu : ECL, H9, salle TIC

Scalable Controllability of Large-Scale Networks: an Output Controllability Approach

In this presentation we consider the problem of controllability and energy consumption for large-scale networks. In this context, computing the (conventional) Controllability Gramian presents computational issues, noise and ill-conditionement. Instead of controlling separately all the nodes of the network, we consider the case in which an output, defined as some measurement (for instance the average) of the nodes which are not directly controlled, needs to be steered to a certain value while minimising the control energy. The concept of Output Controllability and in particular the Output Controllability Gramian is thus exploited to analyze the properties of the system. In this context, we show that it is possible to obtain a reduced-order model which makes the Gramian compution and control design much easier. Simulations show that the reduced model is consistent with the original one and for low ratios of controlled nodes, more robust and performing with respect to the original.

26/04/2018 : Stéphane Marinesco (Bio-Ing)

Lieu : ECL, H9, salle Bourbonnais

Stéphane Marinesco est chargé de recherche (INSERM) au Centre de Recherche en Neurosciences de Lyon (Inserm U1028 - CNRS UMR 5292 - Université Lyon 1). Son exposé est intitulé :
Microbiocapteurs électrochimiques pour l’analyse in vivo du milieu interstitiel

29/03/2018, 13h30-16h15 : présentations de leurs sujets par les 6 doctorants 1A relevant de la Priorité T1 "Systèmes et Energies Sûrs".

  • G. DANTAS DE FREITAS : Comparaison des stratégies de protection (E. Niel / B. Raison, G2ELab)
  • F. ERRIGO : Convertisseurs de puissance avec stockage d’énergie intégré pour réseaux haute tension à courant continu (P. Venet, A. Sari)
  • R. A. PENA : Amélioration de la durée de vie des systèmes de stockage d’énergie électrique par optimisation des convertisseurs d’énergie modulaires associés (P. Venet, A. Hijazi)
  • D. ROMERO : Combined heat and power economic dispatch for isolated Microgrids (E. Niel)
  • E. TOURNON : Conception optimale d’un Vélo à Assistance Electrique à architecture hybride série et à base de supercondensateurs (P. Venet, A. Sari)
  • Z. YANG : Détection et traitement des signes de vieillissement d’un transistor MOSFET de puissance en silicium pour application réseau de bord automobile (B. Allard, G. Clerc)

09/03/2018, 12h30-14h00 : Jorge I. AYALA CUEVAS (PhD Club)

ECL, H9, salle TIC

Control design and robustness analysis of a teleoperated robot-tissue interactions system using IQC approach.

In medical robotics, there exist an increasing interest on the development of collaborative systems that allow to improve the skills of surgeons by using robotic devices. One of the most popular modalities are teleoperated systems. Bilateral teleoperation systems allow a human operator to manipulate a remote environment while getting the feeling of being interacting directly with it through force feedback.
The control of these kind of systems has historically presented challenging problems due to the time varying nature of human operator and environment, in addition to an inherent trade-off between transparency and stability. Classical approaches have rely on two-port networks analogy and passivity theory, they allow to analyse stability by skipping the explicit modelling of human and environment. However, the cost of simplifying the problem is the resulting conservatism of the analysis results.
In recent years, some new model-based robust control approaches have been explored for teleoperated systems. This is mainly motivated by the utilisation of integral quadratic constraints (IQC) tools, which allow to capture the main features of time-varying uncertainties and non-linearities and to establish less conservative stability tests posed as LMI optimisation problems.
This presentation will present the main stages and results of my master’s thesis project: system modelling, identification, control design, robustness analysis and experimental implementation.

23/01/2018, 12h30 : Peng WANG (PhD Club)

ECL, H9, salle à préciser

Active vibration control of a specific zone on the flexible structure by using smart materials

This work proposes a methodology that deals with a specific active vibration control problem, which is the vibration attenuation of a specific zone on a flexible beam. A force disturbance in a wide frequency range that contains more than 10 vibration modes is considered and the vibration in the central zone of the beam is reduced. Multiple piezoelectric patches are used as actuator-sensor pairs and their positions are carefully chosen. The mathematical model of the beam-piezo system is first built and then corrected by Grey-box identification technique. Based on the model, a special output signal is constructed whose power represents the vibration energy in the central zone. A new model reduction technique is proposed to reduce the order of the model based on control objective and then H∞ robust control is performed to give a MIMO feedback controller that minimize the power of the constructed signal. The spill-over problem is also eliminated by considering in the controller design the error between the complete model and the reduced model. The effectiveness of the controller is verified by both simulation and experiment.