This dissertation introduces a hierarchical optimization framework for power converter-based energy management systems, with a primary focus on weight minimization. Emphasizing modularity and scalability, the research systematically tackles the challenges in optimizing these systems, addressing complex design variables, couplings, and the integration of heterogeneous models. The study begins with a comparative evaluation of various metaheuristic optimization methods applied to power inductors and converters, including genetic algorithm, particle swarm optimization, and simulated annealing. This is complemented by a global sensitivity analysis using the Morris method to understand the impact of different design variables on the design objectives and constraints in power electronics. Additionally, a thorough evaluation of different modeling methods for key components is conducted, leading to the validation of selected analytical models at the component level through extensive experiments. Further, the research progresses to studies at the converter level, focusing on a weight-optimized design for the thermal management systems for silicon carbide (SiC) MOSFET-based modular converters and the development of a hierarchical digital control system. This stage includes a thorough assessment of the accuracy of small-signal models for modular converters. At this point, the research methodically examines various design constraints, notably thermal considerations and transient responses. This examination is critical in understanding and addressing the specific challenges associated with converter-level design and the implications on system performance. The dissertation then presents a systematic approach where design variables and constraints are intricately managed across different hierarchies. This strategy facilitates the decoupling of subsystem designs within the same hierarchy, simplifying future enhancements to the optimization process. For example, component databases can be expanded effortlessly, and diverse topologies for converters and subsystems can be incorporated without the need to reconfigure the optimization framework. Another notable aspect of this research is the exploration of the scalability of the optimization architecture, demonstrated through design examples. This scalability is pivotal to the framework's effectiveness, enabling it to adapt and evolve alongside technological advancements and changing design requirements. Furthermore, this dissertation delves into the data transmission architecture within the hierarchical optimization framework. This architecture is not only critical for identifying optimal performance measures, but also for conveying detailed design information across all hierarchy levels, from individual components to entire systems. The interrelation between design specifications, constraints, and performance measures is illustrated through practical design examples, showcasing the framework's comprehensive approach. In summary, this dissertation contributes a novel, modular, and scalable hierarchical optimization architecture for the design of power converter-based energy management systems. It offers a comprehensive approach to managing complex design variables and constraints, paving the way for more efficient, adaptable, and cost-effective power system designs. / Doctor of Philosophy / This dissertation introduces an innovative approach to designing energy control systems, inspired by the creativity and adaptability of a Lego game. Central to this concept is a layered design methodology. The journey begins with power components, the fundamental 'Lego bricks'. Each piece is meticulously optimized for compactness, forming the robust foundation of the system. Like connecting individual Lego bricks into a module, these power components come together to form standardized power converters. These converters offer flexibility and scalability, similar to how numerous structures can be built from the same set of Lego pieces. The final layer involves assembling these power converters in order to construct comprehensive energy control systems. This mirrors the process of using Lego subassemblies to build larger, more intricate structures. At this system-level design, the standardized converters are integrated to optimize overall system performance. Key to this dissertation's methodology is an emphasis on modularity and scalability. It enables the creation of diverse energy control systems of varying sizes and functionalities from these fundamental units. The research delves into the intricacies of design variables and constraints, ensuring that each 'Lego piece' contributes optimally to the bigger picture. This includes exploring the scalability of the architecture, allowing it to evolve with technological advancements and design requirements, as well as examining data transmission within the system to ensure efficient data communication across all levels. In essence, this dissertation is about recognizing the potential in the smallest components and understanding their role in the grand scheme of the system. It is akin to playing a masterful game of Lego, where building something greater from small, well-designed parts leads to more efficient, adaptable, and cost-effective energy control system designs. This approach is particularly relevant for applications in transportation systems and renewable energy in remote locations, showcasing the universal applicability of this 'Lego game' to energy management.

Dans cette thèse, nous nous intéressons à la modélisation de mouvements de foules. Nous proposons un modèle microscopique basé sur la théorie des jeux. Chaque individu a une certaine vitesse souhaitée, celle qu'il adopterait en l'absence des autres. Une personne est influencée par certains de ses voisins, pratiquement ceux qu'elle voit devant elle. Une vitesse réelle est considérée comme possible si elle réalise un équilibre de Nash instantané: chaque individu fait son mieux par rapport à un objectif personnel (vitesse souhaitée), en tenant compte du comportement des voisins qui l'influencent. Nous abordons des questions relatives à la modélisation ainsi que les aspects théoriques du problème dans diverses situations, en particulier dans le cas où chaque individu est influencé par tous les autres, et le cas où les relations d'influence entre les individus présentent une structure hiérarchique. Un schéma numérique est développé pour résoudre le problème dans le second cas (modèle hiérarchique) et des simulations numériques sont proposées pour illustrer le comportement du modèle. Les résultats numériques sont confrontés avec des expériences réelles de mouvements de foules pour montrer la capacité du modèle à reproduire certains effets.Nous proposons une version macroscopique du modèle hiérarchique en utilisant les mêmes principes de modélisation au niveau macroscopique, et nous présentons une étude préliminaire des difficultés posées par cette approche.La dernière problématique qu'on aborde dans cette thèse est liée aux cadres flot gradient dans les espaces de Wasserstein aux niveaux continu et discret. Il est connu que l'équation de Fokker-Planck peut s'interpréter comme un flot gradient pour la distance de Wasserstein continue. Nous établissons un lien entre une discrétisation spatiale du type Volume Finis pour l'équation de Fokker-Planck sur une tesselation de Voronoï et les flots gradient sur le réseau sous-jacent, pour une distance de type Wasserstein récemment introduite sur l'espace de mesures portées par les sommets d'un réseaux. / We are interested in the modeling of crowd motion. We propose a microscopic model based on game theoretic principles. Each individual is supposed to have a desired velocity, it is the one he would like to have in the absence of others. We consider that each individual is influenced by some of his neighbors, practically the ones that he sees. A possible actual velocity is an instantaneous Nash equilibrium: each individual does its best with respect to a personal objective (desired velocity), considering the behavior of the neighbors that influence him. We address theoretical and modeling issues in various situations, in particular when each individual is influenced by all the others, and in the case where the influence relations between individuals are hierarchical. We develop a numerical strategy to solve the problem in the second case (hierarchical model) and propose numerical simulations to illustrate the behavior of the model. We confront our numerical results with real experiments and prove the ability of the hierarchical model to reproduce some phenomena.We also propose to write a macroscopic counterpart of the hierarchical model by translating the same modeling principles to the macroscopic level and make the first steps towards writing such model.The last problem tackled in this thesis is related to gradient flow frameworks in the continuous and discrete Wasserstein spaces. It is known that the Fokker-Planck equation can be interpreted as a gradient flow for the continuous Wasserstein distance. We establish a link between some space discretization strategies of the Finite Volume type for the Fokker- Planck equation in general meshes (Voronoï tesselations) and gradient flows on the underlying networks of cells, in the framework of discrete Wasserstein-like distance on graphs recently introduced.

[pt] A parametrização de superfícies triangularizadas, orientadas e sem bordo não se realiza em um único domínio planar e, por vezes, são sensíveis às descontinuidades ora introduzidas entre os múltiplos domínios planares solicitados. Para tanto, um domínio base não planar é exigido com uma estrutura diferenciável, bem como, a parametrização da superfície por este domínio. A principal abordagem desta tese utiliza uma estrutura de multi-triangulação que direciona a simplificação da superfície inicial numa superfície base e propõem uma projeção hierárquica dos vértices iniciais sobre este domínio. A projeção hierárquica é combinada com um sistema de parametrização da superfície base em domínios locais que são relacionados por funções de transição suaves. Como aplicação, o remalhamento semirregular de superfícies triangularizadas converte a superfície inicial, possivelmente irregular, em outra superfície com triangulação semirregular. A qualidade da triangulação e a preservação da forma original são aspectos importantes para o remalhamento e são resultantes da combinação de componentes envolvidas durante a sua construção, como a superfície base que aproxima a superfície inicial sem bordo, o particionamento semirregular desta superfície base e o reposicionamento de sua geometria. Um desafio significativo está no reposicionamento da geometria dos vértices que decorre da parametrização da superfície inicial. A otimização hierárquica realizada nestes domínios locais buscando reduzir as energias de distorção introduzidas pela parametrização, tal como, os atributos da superfície base são fundamentais para a qualidade deste remalhamento. A estrutura hierárquica permite flexibilidade durante a otimização e influencia no tempo de convergência. / [en] The parameterization of triangulated, oriented and free boundary surfaces does not take place in a single planar domain, and sometimes are sensitive to discontinuities introduced between the various planars domains requested. Therefore, a non-planar base domain is required with a differentiable structure, as well as a parameterization of the surface by this domain. The main approach of this thesis uses a multi-triangulation structure that directs the simplification of the initial surface to base surface and proposes a hierarchical projection of the initial vertices on the domain. A hierarchical projection is combined with a parameterization system of the initial surface composed of local domains related by smooth transition maps. As an application, the semi-regular remeshing of triangulated surfaces converts an input surface, possibly irregular, to another surface with semi-regular triangulation. The quality of the triangulation and the shape preserving are important aspects for the remeshing and are resulting from the components merge used during its building, such as the base surface that approximate the initial free boundary surface, refinement of the semi-regular surface and geometric fitting. An important challenge is the geometric fitting that results from the parameterization of the initial surface. The hierarchical optimization carried out in these local domains that minimizes the distortion energies produced by the parameterizations, such as the base surface attributes are fundamental to the quality of this remeshing. The hierarchical structure allows flexibility during optimization and influences the convergence time.

Nous étudions la convergence de systèmes dynamiques vers des équilibres. En particulier, nous nous intéressons à deux types d'équilibres. D'une part, les solutions d'inéquations variationnelles sous contraintes qui interviennent aussi dans la résolution de problèmes d'optimisation hiérarchique. D'autre part l'état stable d'un système dynamique, c'est à dire l'état où l'énergie du système est nulle. Cette thèse est divisée en deux parties principales, chacune focalisée sur la recherche d'un de ces équilibres. Dans la première partie nous étudions une classe d'algorithmes explicite-implicites pour résoudre certaines inéquations variationnelles sous contraintes. Nous introduisons un algorithme proximal-gradient pénalisé, "splitting forward-backward penalty scheme". Ensuite, nous prouvons sa convergence ergodique faible vers un équilibre dans le cas général d'un opérateur maximal monotone, et sa convergence forte vers l'unique équilibre si l'opérateur est de plus fortement monotone. Nous appliquons aussi notre algorithme pour résoudre des problèmes d'optimisation sous contrainte ou hiérarchique dont les fonctions objectif et de pénalisation sont formées d'une partie lisse et d'une autre non lisse. En effet, nous démontrons la convergence faible de l'algorithme vers un optimum hiérarchique lorsque l'opérateur est le sous-différentiel d'une fonction convexe semi-continue inférieurement et propre. Nous généralisons ainsi plusieurs algorithmes connus et nous retrouvons leurs résultats de convergence en affaiblissant les hypothèses utilisées dans nombre d'entre eux.Dans la deuxième partie, nous étudions l'action d'un contrôle interne local sur la stabilisation indirecte d'un système dynamique couplé formé de trois équations d'ondes, le système de Bresse. Sous la condition d'égalité des vitesses de propagation des ondes, nous montrons la stabilité exponentielle du système. En revanche, quand les vitesses sont différentes, nous prouvons sa stabilité polynomiale et nous établissons un nouveau taux de décroissance polynomial de l'énergie. Ceci étend des résultats présents dans la littérature au sens où le contrôle est localement distribué (et non pas appliqué à tout le domaine) et nous améliorons le taux de décroissance polynomial de l'énergie pour des conditions au bord de type Dirichlet et Dirichlet-Neumann. / We study the convergence of dynamical systems towards equilibria. In particular, we are interested in two types of equilibria. On one hand solutions of constrained variational inequations that are also involved in the resolution of hierarchical optimization problems. On the other hand the stable state of a dynamical system, i.e. the state when the energy of the system is zero. The thesis is divided into two parts, each focused on one of these equilibria. In the first part, we study a class of forward-backward algorithms for solving constrained variational inequalities. We consider a splitting forward-backward penalty scheme. We prove the weak ergodic convergence of the algorithm to an equilibrium for a general maximal monotone operator, and the strong convergence to the unique equilibrium if the operator is an addition strongly monotone. We also apply our algorithm for solving constrained or hierarchical optimization problems whose objective and penalization functions are formed of a smooth and a non-smooth part. In fact, we show the weak convergence to a hierarchical optimum when the operator is the subdifferential of a closed convex proper function. We then generalize several known algorithms and we find their convergence results by weakening assumptions used in a number of them. In the second part, we study the action of a locally internal dissipation law in the stabilization of a linear dynamical system coupling three wave equations, the Bresse system. Under the equal speed wave propagation condition we show that the system is exponentially stable. Otherwise, when the speeds are different, we prove the polynomial stability and establish a new polynomial energy decay rate. This extends results presented in the literature in the sense that the dissipation law is locally distributed (and not applied in the whole domain) and we improve the polynomial energy decay rate with both types of boundary conditions, Dirichlet and Dirichlet-Neumann.