Dans le domaine de l’ingénierie (par exemple l’aéronautique, l’automobile, la biologie, les circuits), les systèmes dynamiques sont le cadre de base utilisé pour modéliser, contrôler et analyser une grande variété de systèmes et de phénomènes. En raison de l’utilisation croissante de logiciels dédiés de modélisation par ordinateur, la simulation numérique devient de plus en plus utilisée pour simuler un système ou un phénomène complexe et raccourcir le temps de développement et le coût. Cependant, le besoin d’une précision de modèle améliorée conduit inévitablement à un nombre croissant de variables et de ressources à gérer au prix d’un coût numérique élevé. Cette contrepartie justifie la réduction du modèle. Pour les systèmes linéaires invariant dans le temps, plusieurs approches de réduction de modèle ont été effectivement développées depuis les années 60. Parmi celles-ci, les méthodes basées sur l’interpolation se distinguent par leur souplesse et leur faible coût de calcul, ce qui en fait un candidat prédestiné à la réduction de systèmes véritablement à grande échelle. Les progrès récents démontrent des façons de trouver des paramètres de réduction qui minimisent localement la norme H2 de l’erreur d’incompatibilité. En général, une approximation d’ordre réduit est considérée comme un modèle de dimension finie. Cette représentation est assez générale et une large gamme de systèmes dynamiques linéaires peut être convertie sous cette forme, du moins en principe. Cependant, dans certains cas, il peut être plus pertinent de trouver des modèles à ordre réduit ayant des structures plus complexes. A titre d’exemple, certains systèmes de phénomènes de transport ont leurs valeurs singulières Hankel qui se décomposent très lentement et ne sont pas facilement approchées par un modèle de dimension finie. En outre, pour certaines applications, il est intéressant de disposer d’un modèle structuré d’ordre réduit qui reproduit les comportements physiques. C’est pourquoi, dans cette thèse, les modèles à ordre réduit ayant des structures de retard ont été plus précisément considérés. Ce travail a consisté, d’une part, à développer de nouvelles techniques de réduction de modèle pour des modèles à ordre réduit avec des structures de retard et, d’autre part, à trouver de nouvelles applications d’approximation de modèle. La contribution majeure de cette thèse couvre les sujets d’approximation et inclut plusieurs contributions au domaine de la réduction de modèle. Une attention particulière a été accordée au problème de l’approximation du modèle optimale pour les modèles structurés retardés. À cette fin, de nouveaux résultats théoriques et méthodologiques ont été obtenus et appliqués avec succès aux repères académiques et industriels. De plus, la dernière partie de ce manuscrit est consacrée à l’analyse de la stabilité des systèmes retardés par des méthodes interpolatoires. Certaines déclarations théoriques ainsi qu’une heuristique sont développées permettant d’estimer de manière rapide et précise les diagrammes de stabilité de ces systèmes. / In the engineering area (e.g. aerospace, automotive, biology, circuits), dynamical systems are the basic framework used for modeling, controlling and analyzing a large variety of systems and phenomena. Due to the increasing use of dedicated computer-based modeling design software, numerical simulation turns to be more and more used to simulate a complex system or phenomenon and shorten both development time and cost. However, the need of an enhanced model accuracy inevitably leads to an increasing number of variables and resources to manage at the price of a high numerical cost. This counterpart is the justification for model reduction. For linear time-invariant systems, several model reduction approaches have been effectively developed since the 60’s. Among these, interpolation-based methods stand out due to their flexibility and low computational cost, making them a predestined candidate in the reduction of truly large-scale systems. Recent advances demonstrate ways to find reduction parameters that locally minimize the H2 norm of the mismatch error. In general, a reduced-order approximation is considered to be a finite dimensional model. This representation is quite general and a wide range of linear dynamical systems can be converted in this form, at least in principle. However, in some cases, it may be more relevant to find reduced-order models having some more complex structures. As an example, some transport phenomena systems have their Hankel singular values which decay very slowly and are not easily approximated by a finite dimensional model. In addition, for some applications, it is valuable to have a structured reduced-order model which reproduces the physical behaviors. That is why, in this thesis, reduced-order models having delay structures have been more specifically considered. This work has focused, on the one hand, in developing new model reduction techniques for reduced order models having delay structures, and, on the other hand, in finding new applications of model approximation. The major contribution of this thesis covers approximation topics and includes several contributions to the area of model reduction. A special attention was given to the H2 optimal model approximation problem for delayed structured models. For this purpose, some new theoretical and methodological results were derived and successfully applied to both academic and industrial benchmarks. In addition, the last part of this manuscript is dedicated to the analysis of time-delayed systems stability using interpolatory methods. Some theoretical statements as well as an heuristic are developed enabling to estimate in a fast and accurate way the stability charts of those systems.

Cette thèse porte sur la modélisation et l’analyse de stabilité de certains mécanismes biologiques complexes en rapport avec le cancer. Un intérêt particulier est porté au cas de l’hématopoïèse et de la leucémie aiguë myéloblastique (LAM). Les modèles utilisés et/ou introduits dans cette thèse se décrivent par des équations aux dérivées partielles structurées en âge, qui se réduisent à des systèmes à retards de plusieurs types (retards ponctuels ou distribués, à support fini ou infini). Ces modèles à retards sont parfois couplés à des équations aux différences, et possiblement avec des paramètres variant dans le temps. Un des principaux challenges dans ce travail consiste à développer des méthodes temporelles, qui se basent sur la construction de fonctionnelles de Lyapunov-Krasovskii strictes, pour les systèmes non-linéaires à retards étudiés. Les principales notions abordées dans ces travaux incluent : l’analyse de stabilité/stabilisation et de robustesse, l’emploi de techniques de modélisation des populations cellulaires saines et malades, l’étude de différentes classes de systèmes dynamiques, (possiblement à temps variant ou à commutation), et également l’introduction de quelques outils issus de l’intelligence artificielle (planification et recherche de solution) dans un contexte de modèles biologiques. Ainsi, les méthodes de modélisation et d’analyse employées dans ce travail ont permis d’une part d’étendre les résultats de stabilité de cette classe de systèmes biologiques, et d’autre part de mieux comprendre certains mécanismes biologiques liés au cancer et sa thérapie. Plus précisément, certains concepts récemment établis en biologie et en médecine sont mis en évidence dans ce travail pour la première fois dans cette classe de systèmes, telles que : la dédifférenciation des cellules (plasticité), ou encore la dormance des cellules cancéreuses dans des modèles tenant compte de la cohabitation entre cellules saines et mutées. Les résultats obtenus sont interprétés dans le cas de l’hématopoïèse et de la LAM, mais ce travail s’applique également à d’autres types de tissus où le cycle cellulaire se produit de façon similaire. / Medical research is looking for new combined targeted therapies against cancer. Our research project -which involves intensive collaboration with hematologists from Saint-Antoine Hospital in Paris- is imbued within a similar spirit and fits the expectations of a better understanding of the behavior of blood cell dynamics. In fact, hematopoiesis provides a paradigm for studying all the mammalian stem cells, as well as all the mechanisms involved in the cell cycle. We address multiple issues related to the modeling and analysis of the cell cycle, with particular insights into the hematopoietic systems. Stability features of the models are highlighted, since systems trajectories reflect the most prominent healthy or unhealthy behaviors of the biological process under study. We indeed perform stability analysis of systems describing healthy and unhealthy situations, with a particular interest in the case of acute myeloblastic leukemia (AML). Thus, we pursue the objectives of understanding the interactions between the various parameters and functions involved in the mechanisms of interest. For that purpose, an advanced stability analysis of the cell fate evolution in treated or untreated leukemia is performed in several modeling frameworks, and our study suggests new anti-leukemic combined chemotherapy. Throughout the thesis, we cover many biological evidences that are currently undergoing intensive biological research, such as: cell plasticity, mutations accumulation, cohabitation between ordinary and mutated cells, control or eradication of cancer cells, cancer dormancy, etc.Among the contributions of Part I of the thesis, we can mention the extension of both modeling and analysis aspects in order to take into account a proliferating phase in which most of the cells may divide, or die, while few of them may be arrested during their cycle for unlimited time. We also introduce for the first time cell-plasticity features to the class of systems that we are focusing on.Next, in Part II, stability analyses of some differential-difference cell population models are performed through several time-domain techniques, including tools of Comparative and Positive Systems approaches. Then, a new age-structured model describing the coexistence between cancer and ordinary stem cells is introduced. This model is transformed into a nonlinear time-delay system that describes the dynamics of healthy cells, coupled to a nonlinear differential-difference system governing the dynamics of unhealthy cells. The main features of the coupled system are highlighted and an advanced stability analysis of several coexisting steady states is performed through a Lyapunov-like approach for descriptor-type systems. We pursue an analysis that provides a theoretical treatment framework following different medical orientations, among which: i) the case where therapy aims to eradicate cancer cells while preserving healthy ones, and ii) a less demanding, more realistic, scenario that consists in maintaining healthy and unhealthy cells in a controlled stable dormancy steady-state. Mainly, sufficient conditions for the regional exponential stability, estimate of the decay rate of the solutions, and subsets of the basins of attraction of the steady states of interest are provided. Biological interpretations and therapeutic strategies in light of emerging AML-drugs are discussed according to our findings.Finally, in Part III, an original formulation of what can be interpreted as a stabilization issue of population cell dynamics through artificial intelligence planning tools is provided. In that framework, an optimal solution is discovered via planning and scheduling algorithms. For unhealthy hematopoiesis, we address the treatment issue through multiple drug infusions. In that case, we determine the best therapeutic strategy that restores normal blood count as in an ordinary hematopoietic system.

No description available.

Le thème principal de ce mémoire est la commande sous contraintes pour des systèmes à retard, en tenant compte de la problématique d’échantillonnage (où les informations concernant le système en temps continu sont, par exemple, envoyées par un réseau de communication) et de la présence de contraintes sur les trajectoires du système et sur l’entrée de commande. Pendant le processus d’échantillonnage, le retard variable dans le temps peut être traité comme une incertitude, le but étant de confiner cette variation dans un polytope, de façon à couvrir toutes les variations possibles du retard. Pour stabiliser des systèmes à retard, nous nous sommes basés sur la théorie de Lyapunov. En utilisant cette méthode, nous pouvons trouver un retour d’état qui stabilise le système malgré la présence du retard variable dans la boucle. Une autre possibilité est l’utilisation des candidates de Lyapunov-Krasovskii. La théorie des ensembles invariants est largement utilisée dans ce manuscrit, car il est souhaitable d’obtenir une région de ≪ sûreté ≫, ou le comportement du système est connu, en dépit de la présence du retard (variable) et des contraintes sur les trajectoires du système. Lorsqu’ils sont obtenus dans l’espace d’état augmenté, les ensembles invariants sont très complexes, car la dimension de l’espace Euclidien sera proportionnelle à la taille du système mais aussi à la taille du retard. Le concept de D-invariance est ainsi proposé. La commande prédictive (en anglais MPC) est présentée, pour tenir compte des contraintes sur les trajectoires et appliquer une commande optimale à l’entrée du système. / The main interest of the present thesis is the constrained control of time-delay system, more specifically taking into consideration the discretization problem (due to, for example, a communication network) and the presence of constraints in the system’s trajectories and control inputs. The effects of data-sampling and modeling problem are studied in detail, where an uncertainty is added into the system due to additional effect of the discretization and delay. The delay variation with respect to the sampling instants is characterized by a polytopic supra-approximation of the discretization/delay induced uncertainty. Some stabilizing techniques, based on Lyapunov’s theory, are then derived for the unconstrained case. Lyapunov-Krasovskii candidates were also used to obtain LMI conditions for a state feedback, in the ``original” state-space of the system. For the constrained control purposes, the set invariance theory is used intensively, in order to obtain a region where the system is ``well-behaviored”, despite the presence of constraints and (time-varying) delay. Due to the high complexity of the maximal delayed state admissible set obtained in the augmented state-space approach, in the present manuscript we proposed the concept of set invariance in the ``original” state-space of the system, called D-invariance. Finally, in the las part of the thesis, the MPC scheme is presented, in order to take into account the constraints and the optimality of the control solution.

The purpose of this thesis is to develop new stability conditions for several linear dynamic systems, including linear parameter-varying (LPV), time-delay systems (LPVTD), slow LPV systems, and parameter-dependent linear time invariant (LTI) systems. These stability conditions are less conservative and/or computationally easier to apply than existing ones. This dissertation is composed of four parts. In the first part of this thesis, the complete stability domain for LTI parameter-dependent (LTIPD) systems is synthesized by extending existing results in the literature. This domain is calculated through a guardian map which involves the determinant of the Kronecker sum of a matrix with itself. The stability domain is synthesized for both single- and multi-parameter dependent LTI systems. The single-parameter case is easily computable, whereas the multi-parameter case is more involved. The determinant of the bialternate sum of a matrix with itself is also exploited to reduce the computational complexity. In the second part of the thesis, a class of parameter-dependent Lyapunov functions is proposed, which can be used to assess the stability properties of single-parameter LTIPD systems in a non-conservative manner. It is shown that stability of LTIPD systems is equivalent to the existence of a Lyapunov function of a polynomial type (in terms of the parameter) of known, bounded degree satisfying two matrix inequalities. The bound of polynomial degree of the Lyapunov functions is then reduced by taking advantage of the fact that the Lyapunov matrices are symmetric. If the matrix multiplying the parameter is not full rank, the polynomial order can be reduced even further. It is also shown that checking the feasibility of these matrix inequalities over a compact set can be cast as a convex optimization problem. Such Lyapunov functions and stability conditions for affine single-parameter LTIPD systems are then generalized to single-parameter polynomially-dependent LTIPD systems and affine multi-parameter LTIPD systems. The third part of the thesis provides one of the first attempts to derive computationally tractable criteria for analyzing the stability of LPV time-delayed systems. It presents both delay-independent and delay-dependent stability conditions, which are derived using appropriately selected Lyapunov-Krasovskii functionals. According to the system parameter dependence, these functionals can be selected to obtain increasingly non-conservative results. Gridding techniques may be used to cast these tests as Linear Matrix Inequalities (LMI's). In cases when the system matrices depend affinely or quadratically on the parameter, gridding may be avoided. These LMI's can be solved efficiently using available software. A numerical example of a time-delayed system motivated by a metal removal process is used to demonstrate the theoretical results. In the last part of the thesis, topics for future investigation are proposed. Among the most interesting avenues for research in this context, it is proposed to extend the existing stability analysis results to controller synthesis, which will be based on the same Lyapunov functions used to derive the nonconservative stability conditions. While designing the dynamic ontroller for linear and parameter-dependent systems, it is desired to take the advantage of the rank deficiency of the system matrix multiplying the parameter such that the controller is of lower dimension, or rank deficient without sacrificing the performance of closed-loop systems.

Cette thèse développe les lois de coordination de systèmes de Lagrange. Elle propose en premier lieu une stratégie complètement décentralisée qui se base sur la technique de cross-coupling pour la commande d'un groupe de robots, appelé réseau, qui synchronisent leurs mouvements en suivant une trajectoire désirée. Cette stratégie est étendue pour faire face à l'incertitude paramétrique des robots ainsi qu’aux retards fréquemment rencontrés dans les applications pratiques de réseaux de communication. Une deuxième architecture basée sur la théorie des graphes est proposée pour les réseaux à leader. L'approche développée est considérée hybride. Une extension adaptative à base de réseaux de neurones est développée pour traiter les cas d'incertitude paramétrique. La stratégie conçue prend en considération les délais dans la réception des données. En se basant sur la notion de système en chaîne, la théorie des graphes, le concept de la passivité et la technique du backstepping, une nouvelle méthodologie de la conception de contrôleur de synchronisation pour une classe de systèmes sous-actionnés est développée. Afin d’avoir la possibilité d’implémenter ces stratégies de contrôle, on a développé une plate-forme d'expérimentation pour la robotique industrielle coopérative. / This thesis investigates the issue of designing decentralized control laws to cooperatively control a team of robot manipulators. The purpose is to synchronize their movements while tracking common desired trajectory. Based on a combination of Lyapunov direct method and cross-coupling technique, To account for unmatched uncertainties, the proposed decentralized control laws are extended to an adaptive synchronization tracking controllers. Moreover, due to communication imperfection, time delay communication problems are considered in the performance analysis of the controllers. Another relevant problem for distributed synchronized systems is the leader-follower control problem. In this strategy, a decentralized control laws based on the backstepping scheme is proposed to deal with a leader-follower multiple robots structure. Based on graph theory, the coordination strategy combines the leader follower control with the decentralized control. The thesis, also considers the cooperative movement of under- actuated manipulators tracking reference trajectories defined by the user. The control problem for a network of class of under-actuated systems is considered. The approach we adopted in this thesis consists in decomposing the under-actuated manipulators into a cascade of passive subsystems that synchronize with he other neighbors subsystems. The resulting synchronized control law is basically a combination of non-regular backstepping procedure aided with some concepts from graph theory. The proposed controllers are validated numerically, assuming that the underlying communication graph is strongly connected. To implement these control strategies, we developed an experimental platform made of three robot manipulators.

Les applications du calcul fractionnaire en automatique se sont considérablement développées ces dernières années, surtout en commande robuste. Ce mémoire est une contribution à la commande robuste des systèmes d'ordre entier à l'aide d'un correcteur PID d'ordre fractionnaire.Le conventionnel régulateur PID, unanimement apprécié pour le contrôle des processus industriels, a été adapté au cas fractionnaire sous la forme PInDf grâce à l'introduction d'un modèle de référence d'ordre non entier, réputé pour sa robustesse vis-à-vis des variations du gain statique.Cette nouvelle structure a été étendue aux systèmes à retard sous la forme d'un Prédicteur de SMITH fractionnaire. Dans leur forme standard, ces correcteurs sont adaptés à la commande des systèmes du premier et du second ordre, avec ou sans retard pur.Pour des systèmes plus complexes, deux méthodologies de synthèse du correcteur ont été proposées, grâce à la méthode des moments et à l'approche retour de sortie.Pour les systèmes dont le modèle est obtenu à partir d'une identification, la boucle fermée doit en outre être robuste aux erreurs d'estimation. Un modèle pire-cas, déduit de la matrice de covariance de l'estimateur et des domaines d'incertitudes fréquentielles, a été proposé pour la synthèse du correcteur.Les différentes simulations numériques montrent l'efficacité de cette méthodologie pour l'obtention d'une boucle fermée robuste aux variations du gain statique et aux incertitudes d'identification. / The application of fractional calculus in automatic control have received much attention these last years, mainly in robust control. This PhD dissertation is a contribution to the control of integer order systems using a fractional order PID controller.The classical PID, well known for its applications to industrial plants, has been adapted to the fractional case as a PInDf controller, thanks to a fractional order reference model, characterized by its robustness to static gain variations.This new controller has been generalized to time delay systems as a fractional SMITH Predictor. In standard case, these controllers are adapted to first and second order systems, with or without a time delay. For more complex systems, two design methodologies have been proposed, based on the method of moments and on output feedback approach.For systems whose model is obtained by an identification procedure, the closed loop has to be robust to estimation errors. So, a worst-case model, derived from the covariance matrix of the estimator and the frequency uncertainty domains, has been proposed for the design of the controller.The different numerical simulations demonstrate that this methodology is able to provide robustness to static gain variations and to identification uncertainties.

Nesta Dissertação são propostos dois esquemas de controle para sistemas não-lineares com atraso. No primeiro, o objetivo é controlar uma classe de sistemas incertos multivariáveis, de grau relativo unitário, com perturbações não-lineares descasadas dependentes do estado, e com atraso incerto e variante no tempo em relação ao estado. No segundo, deseja-se controlar uma classe de sistemas monovariáveis, com parâmetros conhecidos, grau relativo arbitrário, atraso arbitrário conhecido e constante na saída. Admitindo-se que o atraso na entrada pode ser deslocado para a saída, então, o segundo esquema de controle pode ser aplicado a sistemas com atraso na entrada. Os controladores desenvolvidos são baseados no controle por modo deslizante e realimentação de saída, com função de modulação para a amplitude do sinal de controle. Além disso, observadores estimam as variáveis de estado não-medidas. Em ambos os esquemas de controle propostos, garante-se propriedades de estabilidade globais do sistema em malha fechada. Simulações ilustram a eficácia dos controladores desenvolvidos. / Two control schemes for nonlinear time-delay systems are proposed in this thesis. The purpose of the first scheme is to control a class of uncertain multivariable systems, with relative degree one, nonlinear unmatched state dependent disturbances, and uncertain time-varying state delay. The purpose of the second scheme is to control a class of single-input-single-output systems, with known parameters, arbitrary relative degree, with constant and known arbitrary output delay. Assuming that input delays can be transferred to the output, so the second scheme can be applied to systems with input time-delay. The developed controllers are based on sliding mode control and output feedback, with modulation function to the control signal amplitude. Furthermore, observers estimate unmeasured state variables. In both schemes, global stability properties of the closed loop system are guaranteed. Simulations illustrate the effectiveness of the proposed approaches.

Thesis submitted in fulfilment of the requirements for the degree Doctor of Technology: Electrical Engineering in the Faculty of Engineering at the Cape Peninsula University of Technology, 2013 / Networked Control Systems is a control system where the plant and the controller exchange information via a shared communication network and the network is considered as part of the closed loop control system. Unfortunately the network introduces network induced random varying time delays and data packet loss amongst the communication network imperfections. The network delays are considered to be between the controller and the actuator and between the sensor and the controller. These network imperfections degrade the performance of the closed loop control system and result in closed loop system instability. The complexity of measuring the communication network imperfection in networked control systems makes it difficult for the control engineers to develop methods for design of controllers that can incorporate and compensate these imperfections in order to improve the performance of the networked control systems. In this thesis a co-simulation toolset called LabNS2 is developed to address the first problem of measuring the communication network imperfections by providing an ideal environment that can be used to investigate the influence of network time delays or packet loss. The software environment of the toolset is based on LabVIEWTM and Network Simulator Version 2 (NS2). A new robust predictive optimal controller design method is developed to address the problem of the destabilising effect of the network induced time delay between the controller and the actuator. The design approach is based on time shifting of the optimisation horizon and a state predictor. The design of the controller is based on a model of the plant with delay in the control vector equal to the delay between the controller and the actuator or to the sum of the delays between the controller and the actuator and between the sensor and the controller. The time shifting approach allows the design of the controller to be performed for a model without time delay. Then the control action is based on the future values of the state space vector estimates. The state predictor is developed to predict these future values of the state using the present and past values of the state estimates and control actions. This technique is made possible by the use of the plant model Transition Matrix. A Discrete Kalman Filter is modified to address the problem of the destabilising effect of the network induced time delay between the sensor and the controller. An additional state estimation vector is added to the filter estimate at every current moment of time. iv The developed methods are implemented for networked control of a dish antenna driven by two stepper motors. The outcomes of the thesis can be used for the education and fundamental research purposes, but the developed control strategies have significant sense towards the Square Kilometer Array projects and satellite systems industry. / National Research Foundation

Nesta Dissertação são propostos dois esquemas de controle para sistemas não-lineares com atraso. No primeiro, o objetivo é controlar uma classe de sistemas incertos multivariáveis, de grau relativo unitário, com perturbações não-lineares descasadas dependentes do estado, e com atraso incerto e variante no tempo em relação ao estado. No segundo, deseja-se controlar uma classe de sistemas monovariáveis, com parâmetros conhecidos, grau relativo arbitrário, atraso arbitrário conhecido e constante na saída. Admitindo-se que o atraso na entrada pode ser deslocado para a saída, então, o segundo esquema de controle pode ser aplicado a sistemas com atraso na entrada. Os controladores desenvolvidos são baseados no controle por modo deslizante e realimentação de saída, com função de modulação para a amplitude do sinal de controle. Além disso, observadores estimam as variáveis de estado não-medidas. Em ambos os esquemas de controle propostos, garante-se propriedades de estabilidade globais do sistema em malha fechada. Simulações ilustram a eficácia dos controladores desenvolvidos. / Two control schemes for nonlinear time-delay systems are proposed in this thesis. The purpose of the first scheme is to control a class of uncertain multivariable systems, with relative degree one, nonlinear unmatched state dependent disturbances, and uncertain time-varying state delay. The purpose of the second scheme is to control a class of single-input-single-output systems, with known parameters, arbitrary relative degree, with constant and known arbitrary output delay. Assuming that input delays can be transferred to the output, so the second scheme can be applied to systems with input time-delay. The developed controllers are based on sliding mode control and output feedback, with modulation function to the control signal amplitude. Furthermore, observers estimate unmeasured state variables. In both schemes, global stability properties of the closed loop system are guaranteed. Simulations illustrate the effectiveness of the proposed approaches.