Past research has shown that natural ventilation can be used to satisfy upwards of 98% of the yearly cooling demand when utilized in the appropriate climate zone. Yet widespread implementation of natural ventilation has been limited in practice. This delay in market adoption is mainly due to lack of effective and reliable control. Historically, control of natural ventilation was left to the occupant (i.e. they are responsible for opening and closing their windows) because occupants are more readily satisfied when given control of the indoor environment. This strategy has been shown to be effective during summer months, but can lead to both over and under ventilation, as well as the associated unnecessary energy waste during the winter months. This research presents the development and evaluation of a model-based control algorithm for natural ventilation. The proposed controller is designed to modulate the operable windows based on ambient temperature, wind speed, wind direction, solar radiation, indoor temperature and other building characteristics to ensure adequate ventilation and thermal comfort throughout the year without the use of mechanical ventilation and cooling systems. A midrise student dormitory building, located in Portland OR, has been used to demonstrate the performance of the proposed controller. Simulation results show that the model-based controller is able to reduce under-ventilated hours to 6.2% of the summer season (June - September) and 2.5% of the winter (October - May) while preventing over-heating during 99% of the year. In addition, the model-based-controller reduces the yearly energy cost by 33% when compared to a conventional heat pump system. As a proactive control, model-based control has been used in a wide range of building control applications. This research serves as proof-of-concept that it can be used to control operable windows to provide adequate ventilation year-round without significantly affecting thermal comfort. The resulting control algorithm significantly improves the reliability of natural ventilation design and could lead to a wider adoption of natural ventilation in appropriate climate zones.

Les exigences croissantes sur l'efficacité énergétique des bâtiments, l'évolution du {marché} énergétique, le développement technique récent ainsi que les particularités du poste de chauffage ont fait du MPC le meilleur candidat pour la régulation thermique des bâtiments à occupation intermittente. Cette thèse présente une méthodologie basée sur la commande prédictive distribuée visant un compromis entre l'optimalité, la simplicité et la flexibilité de l'implantation de la solution proposée. Le développement de l'approche est progressif : à partir du cas d'une seule zone, la démarche est ensuite étendue au cas multizone et / ou multisource, avec la prise en compte des couplages thermiques entre les zones adjacentes. Après une formulation quadratique du critère MPC pour mieux satisfaire les objectifs économiques du contrôle, la formulation linéaire est retenue. Pour répartir la charge de calcul, des méthodes de décomposition linéaire (comme Dantzig-Wolfe et Benders) sont employées. L'efficacité des algorithmes distribués proposés est illustrée par diverses simulations. / The increasing requirements on energy efficiency of buildings, the evolution of the energy market, the technical developments and the characteristics of the heating systems made of MPC the best candidate for thermal control of intermittently occupied buildings. This thesis presents a methodology based on distributed model predictive control, aiming a compromise between optimality, on the one hand, and simplicity and flexibility of the implementation of the proposed solution, on the other hand. The development of the approach is gradually. The mono-zone case is initially considered, then the basic ideas of the solution are extended to the multi-zone and / or multi-source case, including the thermal coupling between adjacent zones. Firstly we consider the quadratic formulation of the MPC cost function, then we pass towards a linear criterion, in order to better satisfy the economic control objectives. Thus, linear decomposition methods (such as Dantzig-Wolfe and Benders) represent the mathematical tools used to distribute the computational charge among the local controllers. The efficiency of the distributed algorithms is illustrated by simulations.

Cette thèse porte sur l’application des approches de commande prédictive pour l’optimisation des démarrages des centrales à cycles combinés. Il s’agit d’une problématique à fort enjeu qui pose des défis importants. L’élaboration des approches est progressive. Dans une première partie un modèle de centrale est construit et adapté à l’optimisation, en utilisant une méthodologie qui transforme des modèles physiques Modelica conçus pour la simulation en des modèles pour l’optimisation. Cette méthodologie a permis de construire une bibliothèque adaptée à l’optimisation. La suite des travaux porte sur l’utilisation du modèle afin d’optimiser phase par phase les performances du démarrage. La solution proposée optimise, en temps continu, le profil de charge des turbines en recherchant dans des ensembles de fonctions particulières. Le profil optimal est déterminé en considérant que celui-ci peut être décrit par une fonction paramétrée dont les paramètres sont calculés en résolvant un problème de commande optimale sous contraintes. La dernière partie des travaux consiste à intégrer cette démarche d’optimisation à temps continu dans une stratégie de commande à horizon glissant. Cette approche permet d’une part de corriger les dérives liées aux erreurs de modèles et aux perturbations, et d’autre part, d’améliorer le compromis entre le temps de calcul et l’optimalité de la solution. Cette approche de commande conduit cependant à des temps de calcul importants. Afin de réduire le temps de calcul, une structure de commande prédictive hiérarchisée avec deux niveaux, en travaillant à des échelles de temps et sur des horizons différents, a été proposée. / This thesis focuses on the application of model predictive control approaches to optimize the combined cycle power plants start-ups. Generally, the optimization of start-up is a very problematic issue that poses significant challenges. The development of the proposed approaches is progressive. In the first part a physical model of plant is developed and adapted to optimization purposes, by using a methodology which transforms Modelica model components into optimization-oriented models. By applying this methodology, a library suitable for optimization purposes has been built.In the second part, based on the developed model, an optimization procedure to improve the performances of the start-up phases is suggested. The proposed solution optimizes, in continuous time, the load profile of the turbines, by seeking in specific sets of functions. The optimal profile is derived by considering that this profile can be described by a parameterized function whose parameters are computed by solving a constrained optimal control problem. In the last part, the open-loop optimization procedure has been integrated into a receding horizon control strategy. This strategy represents a robust solution against perturbation and models errors, and enables to improve the trade-off between computation time and optimality of the solution. Nevertheless, the control approach leads to a significant computation time. In order to obtain real-time implementable results, a hierarchical model predictive control structure with two layers, working at different time scales and over different prediction horizons, has been proposed.

This thesis describes the theory of model predictive control and application of the theory to synchronous drives. It shows explicit and on-line solutions and compares the results with classical vector control structure.

L'agriculture est un secteur d'activité qui est confronté aujourd'hui à des objectifs d'accroissement de productivité pour subvenir aux besoins alimentaires de la population mondiale en pleine explosion démographique. Cependant, cette activité sollicite fortement les biens environnementaux tels que l'eau et le sol, et des solutions sont aujourd'hui recherchées pour limiter l'incidence des pratiques agricoles sur l'environnement. Les systèmes de guidage des véhicules agricoles font partie de ces nouvelles technologies qui contribuent à cet objectif, en offrant la possibilité d'assurer la précision du suivi des trajectoires dans les parcelles, et de favoriser ainsi l'efficacité et la qualité du travail agronomique réalisé. Des fonctionnalités essentielles font néanmoins aujourd'hui défaut à ces systèmes. Citons la capacité à compenser la marche en crabe du véhicule sur les terrains glissants en pente, la capacité à contrôler les trajectoires des outils agricoles traînés, et la capacité à effectuer certaines manoeuvres en zone de fourrière. Ce travail de thèse aborde l'ensemble de ces problématiques au travers l'étude de la commande en milieu naturel de robot mobile poly-articulé à roues (RMPA), composé d'un véhicule " tracteur " à deux trains directeurs associé à n remorques passives à attache déportée. Une modélisation cinématique étendue est d'abord adoptée pour tenir compte des effets induits par les faibles conditions d'adhérence sur le comportement global du RMPA. Les variables de glissement introduites sur chacun des trains directeurs et roulants sont estimées à l'aide d'un observateur bâti à la manière d'une loi de commande. La trajectoire de référence à suivre gamma est quant à elle préalablement apprise ou construite à l'aide de primitives élémentaires et d'arcs de clothoïdes pour générer les manoeuvres de demi-tour. En premier lieu, les deux trains directeurs du RMPA sont exploités pour contrôler avec précision non seulement l'écart latéral mais également l'écart angulaire du véhicule " tracteur " par rapport à gamma : la commande du train directeur avant est basée sur la transformation du modèle en un système chaîné, conduisant à un découplage exact des performances latérales et longitudinales, puis sur des techniques de linéarisation exacte pour assurer la régulation latérale. La commande du train directeur arrière se base sur la dynamique de l'écart angulaire pour compenser les glissements et asservir cet écart sur le point de fonctionnement choisi. En second lieu, ces commandes sont étendues pour asservir latéralement la ieme remorque du RMPA le long de gamma : une approche en cascade est utilisée pour traduire une commande virtuelle de la ieme remorque en terme de commande du train directeur avant du véhicule " tracteur ". La commande longitudinale du RMPA est quant à elle basée sur une stratégie de commande prédictive à modèle interne, afin de suivre avec précision le profil de vitesse associé à gamma. De nombreuses expérimentations en conditions réelles, effectuées sur un RMPA composé d'un véhicule " tracteur " à deux trains directeurs et d'une remorque passive à attache déportée, viennent valider les différentes approches présentées dans ce mémoire et permettent d'apprécier les performances des lois de commande proposées. / Agriculture has today the challenge to increase its productivity in order to supply enough food for the growing needs of the world population. However, this activity strongly damages environmental ressources as water and soil, and new solutions are today required to reduce the impact of agricultural practices on environment. Automatic guidance systems for farm vehicles are some new technologies that contribute to this objective, allowing accurate path following in the fields and therefore improving efficiency and quality of the agricultural work carried out. Essential functionalities are however absent in these systems, as the capacities to compensate for the crabway motion of the vehicle on sliding sloping fields, to control the trajectories of the towed implement,and to perform U-turn maneuvers in headland. This thesis studies these problems through the control in presence of sliding of a poly-articulated wheeled mobile robot called RMPA, composed of a four-wheel-steering vehicle and n passive trailers hooked up at some distance from the rear axle of the previous one. An extended kinematic model of the RMPA is first used in order to take into account for the sliding effects on the overall behaviour of the robot. The sliding parameters, introduced on each rolling and steering axle of the RMPA, are estimated using a state observer built as a control law. The reference path [gamma] to be followed is either previously learned or specially planned using elementary primitives connected together with pieces of clothoid to produce the U-turn maneuvers. In a first step, both front and rear steering actuations of the robot are used to accurately control both lateral and angular deviations of the RMPA's four-wheel-steering vehicle with respect to [gamma] : the control of the front steering wheels is based on the transformation of the extended kinematic model into a chained system, allowing to dissociate the lateral and longitudinal commands, and on exact linearization techniques in order to servo lateral deviation. The control of the rear steering wheels is built from the angular deviation dynamic and ensures the convergence of the vehicle deviation to the desired set point. In a second step, these control laws are extended to control the lateral deviation of the ith trailer of the RMPA with respect to [gamma] : a backstepping approach is proposed to calculate the control law for the RMPA's front steering wheels from the study of a virtual control law for the ith trailer. Finally, the longitudinal control law of the RMPA is based on model predictive control, in order to accurately follow the velocity references linked to [gamma]. Numerous experiments relying on an actual RMPA, composed of a passive trailer hooked up at some distance from the rear axle of a four-wheel-steering vehicle, permit to validate the various approaches presented in this thesis and to appreciate the capabilities of the proposed control laws.

A pesquisa em redes neurais artificiais (RNAs) está atualmente experimentando um crescente interesse por modelos que utilizem a variável tempo como um grau de liberdade extra a ser explorado nas representações neurais. Esta ênfase na codificação temporal (temporal coding) tem provocado debates inflamados nas neurociências e áreas correlatas, mas nos últimos anos o surgimento de um grande volume de dados comportamentais e fisiológicos vêm dando suporte ao papel-chave desempenhado por este tipo de representação no cérebro [BALLARD et al. (1998)]. Contribuições ao estudo da representação temporal em redes neurais vêm sendo observadas nos mais diversos tópicos de pesquisa, tais como sistemas dinâmicos não-lineares, redes oscilatórias, redes caóticas, redes com neurônios pulsantes e redes acopladas por pulsos. Como conseqüência, várias tarefas de processamento da informação têm sido investigada via codificação temporal, a saber: classificação de padrões, aprendizagem, memória associativa, controle sensório-motor, identificação de sistemas dinâmicos e robótica. Freqüentemente, porém, não fica muito claro até que ponto a modelagem dos aspectos temporais de uma tarefa contribui para aumentar a capacidade de processamento da informação de modelos neurais. Esta tese busca apresentar, de uma maneira clara e abrangente, os principais conceitos e resultados referentes à proposição de dois modelos de redes neurais não-supervisionadas (RNATs), e como estas lançam mão da codificação temporal para desempenhar melhor a tarefa que lhes é confiada. O primeiro modelo, chamado rede competitiva Hebbiana temporal (competitive temporal Hebbian - CTH), é aplicado especificamente em tarefas de aprendizagem e reprodução de trajetórias do robô manipulador PUMA 560. A rede CTH é uma rede neural competitiva cuja a principal característica é o aprendizado rápido, em apenas uma época de treinamento, de várias trajetórias complexas contendo ) elementos repetidos. As relações temporais da tarefa, representadas pela ordem temporal da trajetória, são capturadas por pesos laterais treinados via aprendizagem hebbiana. As propriedades computacionais da rede CTH são avaliadas através de simulações, bem como através da implementação de um sistema de controle distribuído para o robô PUMA 560 real. O desempenho da rede CTH é superior ao de métodos tabulares (look-up table) tradicionais de aprendizagem de trajetórias robóticas e ao de outras técnicas baseadas em redes neurais, tais como redes recorrentes supervisionadas e modelos de memória associativa bidirecional (BAM). O segundo modelo, chamado rede Auto-Organizável NARX (Self-Organizing NARX-SONARX), é baseado no conhecido algoritmo SOM, proposto por KOHONEN (1997). Do ponto de vista computacional, as propriedades de rede SONARX são avaliadas em diferentes domínios de aplicação, tais como predição de séries temporais caóticas, identificação de um atuador hidráulico e no controle preditivo de uma planta não-linear. Do ponto de vista teórico, demonstra-se que a rede SONARX pode ser utilizada como aproximador assintótico de mapeamentos dinâmicos não-lineares, graças a uma nova técnica de modelagem neural, chamada Memória Associativa Temporal via Quantização Vetorial (MATQV). A MATQV, assim como a aprendizagem hebbiana da rede CTH, é uma técnica de aprendizado associativo temporal. A rede SONARX é comparada com modelos NARX supervisionados, implementados a partir das redes MLP e RBF. Em todos os testes realizados para cada uma das tarefas citadas no parágrafo anterior, a rede SONARX tem desempenho similar ou melhor do que o apresentado por modelos supervisionados tradicionais, com um custo computacional consideravelmente menor. A rede SONARX é também comparada com a rede CTH na parendizagem de trajetórias robóticas complexas, com o intuito de destacar as principais diferenças entre os dois ) tipos de aprendizado associativo. Esta tese também propõe uma taxonomia matemática, baseada na representação por espaço de estados da teoria de sistemas, que visa classificar redes neurais não-supervisionadas temporais com ênfase em suas propriedades computacionais. Esta taxonomia tem como principal objetivo unificar a descrição de modelos neurais dinâmicos, facilitando a análise e a comparação entre diferentes arquiteturas, contrastando suas características representacionais e operacionais. Como exemplo, as redes CTH e a SONARX serão descritas usando a taxonomia proposta. / Neural network research is currently witnessing a significant shift of emphasis towards temporal coding, which uses time as an extra degree of freedom in neural representations. Temporal coding is passionately debated in neuroscience and related fields, but in the last few years a large volume of physiological and behavioral data has emerged that supports a key role for temporal coding in the brain [BALLARD et al. (1998)]. In neural networks, a great deal of research is undertaken under the topics of nonlinear dynamics, oscillatory and chaotic networks, spiking neurons, and pulse-coupled networks. Various information processing tasks are investigated using temporal coding, including pattern classification, learning, associative memory, inference, motor control, dynamical systems identification and control, and robotics. Progress has been made that substantially advances the state-of-the-art of neural computing. In many instances, however, it is unclear whether, and to what extent, the temporal aspects of the models contribute to information processing capabilities. This thesis seeks to present, in a clear and collective way, the main issues and results regarding the proposal of two unsupervised neural models, emphasizing how these networks make use of temporal coding to perform better in the task they are engaged in. The first model, called Competitive Temporal Hebbian (CTH) network, is applied specifically to learning and reproduction of trajectories of a PUMA 560 robot. The CTH model is a competitive neural network whose main characteristic is the fast learning, in just one training epoch, of multiple trajectories containing repeated elements. The temporal relationships within the task, represented by the temporal order of the elements of a given trajectory, are coded in lateral synaptic trained with hebbian learning. The computational properties of the CTH network are assessed through simulations, as well ) as through the practical implementation of a distributed control system for the real PUMA 560 robot. The CTH performs better than conventional look-up table methods for robot trajectory learning, and better than other neural-based techniques, such as supervised recurrent networks and bidirectional associative memory models. The second model, called Self-Organizing NARX (SONARX) network, is based on the well-known SOM algorithm by KOHONEN (1997). From the computational view-point, the properties of the SONARX model are evaluated in different application domains, such as prediction of chaotic time series, identification of an hydraulic actuator and predictive control of a non-linear plant. From the theoretic viewpoint, it is shown that the SONARX model can be seen as an asymptotic approximator for nonlinear dynamical mappings, thanks to a new neural modelling technique, called Vector-Quantized Temporal Associative Memory (VQTAM). This VQTAM, just like the hebbian learning rule of the CTH network, is a temporal associative memory techniques. The SONARX network is compared with supervised NARX models which based on the MLP and RBF networks. For all simulations, in each one of the forementioned application domains, the SONARX network had a similar and sometimes better performance than those observed for standard supervised models, with the additional advantage of a lower computational cost. The SONARX model is also compared with the CTH network in trajectory reproduction tasks, in order to contrast the main differences between these two types of temporal associative learning models. In this thesis, it is also proposed a mathematical taxonomy, based on the state-space representation of dynamical systems, for classification of unsupervised temporal neural networks with emphasis in their computational properties. The main goal of this taxonomy is to unify the description of dynamic neural models, ) facilitating the analysis and comparison of different architectures by constrasting their representational and operational characteristics. Is is shown how the CTH and SONARX models can be described using the proposed taxonomy.

Cette thèse porte sur l'optimisation et la commande prédictive des centrales de production d'énergie en utilisant des modèles physiques des installations. Les modèles sont réalisés à l'aide du langage Modelica, un langage équationnel adapté à la modélisation de systèmes multi-physiques. La modélisation de systèmes physiques dans ce langage est présentée dans une première partie, ainsi que les traitements symboliques réalisés par les compilateurs Modelica pour mettre les modèles sous une forme adaptée à l'optimisation. On présente dans une seconde partie le développement d'une méthode d'optimisation dynamique hybride pour les centrales de production d'énergie, qui fournit une trajectoire optimisée de l'installation sur un horizon long. Les trajectoires calculées incluent les trajectoires des commandes continues ainsi que les décisions d'engagement des différents équipements. L'algorithme d'optimisation combine la méthode de collocation et une méthode nommée Sum Up Rounding (SUR) pour la prise en compte des décisions d'engagement. Un algorithme de commande prédictive (MPC) est enfin introduit afin de garantir le suivi des trajectoires optimales et de prendre en compte en temps réel la présence de perturbations et les erreurs du modèle d'optimisation. L'algorithme MPC utilise des modèles linéarisés tangents générés automatiquement à partir du modèle non linéaire. / This thesis deals with hybrid optimal control and Model Predictive Control (MPC) of power plants by use of physical models. Models of the facilities are developped with Modelica, an equation based language tailored for modelling multi-physics systems. Modeling of physical systems with Modelica is introduced in a first part, as well as some of the symbolic processing done by Modelica compilers that transform the original model to a form suited for optimization. Then, a method to solve optimal control problems on hybrid systems (such as power plants) is presented. This methods provides an optimal trajectory for the power plant on a long horizon. The optimal trajectory computed by the method includes the trajectories of continuous inputs as well as switching decisions for components in the plant. The optimization algorithm combines the collocation method and a method named Sum Up Rounding (SUR) for dealing with switches. Finally, a Model Predictive Controller is developped in order to follow this optimal trajectory in real time, and to cope with disturbances on the actual system and modelling errors. The proposed MPC uses tangent linear models of the plant that are derived automatically from the nonlinear model.

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.

Actuellement la nécessité des systèmes d’alimentation d’énergie, capables d’assurer un fonctionnement stable dans des domaines de fonctionnement assez larges avec des bonnes performances dynamiques (rapidité du système, variations limitées de la tension de sortie en réponse aux perturbations de charge ou de tension d’alimentation), devient de plus en plus importante. De ce fait, cette thèse est orientée sur la commande des convertisseurs de puissance DC-DC représentés par des modèles hybrides.En tenant compte de la structure variable de ces systèmes à commutation, un modèle hybride permet de décrire plus précisément le comportement dynamique d’un convertisseur dans son domaine de fonctionnement. Dans cette optique, l’approximation PWA est utilisée afin de modéliser les convertisseurs DC-DC. A partir des modèles hybrides développés, on s’est intéressé à la stabilisation des convertisseurs au moyen des correcteurs à gains commutés élaborés sur la base de fonctions de Lyapunov PWQ, et à l’implantation d’une commande prédictive explicite, en considérant des contraintes sur l’entrée de commande. La méthode de modélisation et les stratégies de commande proposées ont été appliquées sur deux topologies : un convertisseur buck, afin de mieux maîtriser le réglage des correcteurs et un convertisseur flyback avec filtre d’entrée. Cette dernière topologie nous a permis de répondre aux difficultés du point de vue de la commande (comportement à déphasage non-minimal) rencontrées dans la majorité des convertisseurs DC-DC. Les performances des commandes élaborées ont été validées en simulation sur les topologies considérées et expérimentalement sur une maquette du convertisseur buck. / Lately, power supply systems, guaranteeing the global stability for large enough operation ranges with good dynamic performances (small settling time, bounded overshoot of the output voltage in the presence of load or supply voltage variations), are strongly needed. Therefore, this thesis deals with control problems of DC-DC power converters represented by hybrid models.Considering the variable structure of these switched systems, a hybrid model describes more precisely the converter’s dynamics in its operating domain. From this perspective, a PWA (piecewise affine) approximation is used in order to model the DC-DC converters. Based on the developed hybrid models, first we have designed a stable piecewise linear state-feedback controller using piecewise quadratic (PWQ) Lyapunov functions, and secondly, we have implemented an explicit predictive control law taking into account constraints on the control input. The hybrid modeling technique and the proposed control strategies were applied on two different topologies of converters: a buck converter, in order to have a thorough knowledge of the controllers’ tuning, and a flyback converter with an input filter. This last topology, allowed us to manage different control problems (non-minimum phase behavior) encountered in the majority of topologies of DC-DC power converters. The controllers’ performances were validated in simulation on both considered topologies and also experimentally on buck converter.

Dans cette thèse, on s’intéresse à la commande des robots marcheurs. Contrôler ces systèmes naturellement instables, de dynamique non linéaire, non convexe, de grande dimension, et dépendante des contacts représente un défi majeur en robotique mobile. Les approches classiques formulent une chaîne de contrôle formée d’une cascade de sous problèmes tels que la perception, le planning, la commande du corps complet et l’asservissement articulaire. Les contributions rapportées ici ont toutes pour but d’introduire une rétroaction au niveau de la commande du corps complet ou du planning. Précisément, une première contribution technique est la formulation et la comparaison expérimentale de deux estimateurs de la base du robot. Une seconde contribution est l’implémentation d’un contrôleur par dynamique inverse pour contrôler en couple le robot HRP-2. Une variante de ce contrôleur est aussi formulée et testée en simulation pour stabiliser un robot en contact flexible avec son environnement. Finalement un générateur de marche par commande pré-dictive et couplé à un contrôleur corps complet est présenté. / In this thesis, we are interested in the control of walking robots. Controlling these naturally unstable, non-linear, non-convex, large and contact-dependent systems is a major challenge in mobile robotics. Traditional approaches formulate a chain of control formed by a cascade of sub-problems such as perception, planning, full body control and joint servoing. The contributions reported here are all intended to provide state feedback at the whole body control stage or at the planning stage. Specifically, a first technical contribution is the formulation and experimental comparison of two estimators of the robot base. A second contribution is the implementation of a reverse dynamic controller to control the HRP-2 robot in torque. A variant of this controller is also formulated and tested in simulation to stabilize a robot in flexible contact with its environment. Finally, a predictive control operation generator coupled to a whole body controller is presented.