My thesis deals with possibilities of using evolutionary computation in the field of automation. The theoretical part of the thesis describes the techniques used in the automation and optimization. The practical part of the thesis connects these two disciplines, the output of this work is a program for automatic design of parameters of regulator using a genetic algorithm.

Dans le domaine de l’aéronautique, l’acception ‘shape morphing’ a été utilisée pour identifier ces avions qui subissent certaine modifications géométrique pour améliorer leur adaptation au différents profils de missions. Différemment de la solution classique, celle ‘shape morphing’ exige : une distribution d’actionnement avec une densité de puissance élevée, une mécanisation des structures, des revêtements souples, et le développement des loi de contrôle. Dans ce scénario, un modèle capable reconnaitre l'insertion de capteur et d'actionneur de nouvelle génération, et capable de réduire au minimum le coût du calcul devient très intéressant. Ce travail essaye d'affronter deux aspects différents du problème. Dans la première partie, la question suivante a été exploitée: pour un problème donné, géométrie, chargement, etc. .. condition aux limites, quel est le modèle le plus précis en terme de résultats, fidélité et avec le plus réduit coût de calcul? Deux approches différentes ont été utilisées pour donner une réponse. Le diagramme de la « Théorie Meilleur Plate (RTPB) » a été dessiné. Avec cet instrument il est possible, pour un problème donné, d'identifier les modèles ayants les meilleurs temps de calcul et une bonne fidélité des résultats. Dans la deuxième partie de la thèse un modèle avancé mono dimensionnel et multi domaine en éléments fini est présenté. Le modèle est capable de capturer l'insertion d'éléments piézo-électriques dans l'aile composite. Il a été développé à partir de la formulation Carrera unifiée et à partir de l'équation de comportement électro-mécanique. Une comparaison avec la bibliographie actuelle a été fait afin de valider les résultats. / In the field of aeronautics, shape morphing has been used to identify those aircraft that un-dergo certain geometrical changes to enhance or adapt to their mission profile. Different formthe classical solution the shape morphing required: distributed high-power density actuation, structural mechanization, flexible skins, and control law development. In these scenario, model able to capture the insertion of new generation sensor and actuator, and able to minimize the computational cost become very interesting. These work try to affront two different aspect of the problem. In the first part the following question has been exploited: for a given problem, geometry, loading, boundary condition etc... which is the most accurate model in term of results fidelity with the lowest computational cost? Two different approaches have been used to give an answer. The Best Plate Theory Diagram (BPTD) has been drawn. Trough the BPTD it is possible, for a given problem, to identify those models with the lowest computational time and a good results fidelity. An advanced mono-dimensional multi-field FEM model is presented in the second part of the thesis. The model is able to capture the insertion of piezo-electric elements in composite wing. It has been developed starting from the Carrera Unified Formulation and from the electro-mechanical constitutive equation. Comparison with the bibliography have be done in order to validate the results. / Nel campo dell’aeronautica il termine shape morphing identifica quei velivoli in grado di apportare determinati cambiamenti geometrici al fine di adattarsi a diversi profili di missione. Diversamente dalle soluzioni convenzionali la progettazione di velivoli shape morphing richiede : un’attuazione distribuita, uno skin flessibile in grado di pemettere le deformazioni e delle leggi di controllo. Divengono quindi di notevole interesse modelli in grado di cogliere l’inserzione di attuatori e sensori di nuova generazione all’interno dell’ala, e di esibire al contempo un basto costo computazionale. Nel lavoro presentato in questa tesi vengono trattati entrabi gli aspetti. Nella prima parte si è andati a dare una risposta alla segunete domanda: per un dato problema, geometria, condizioni di carico, etc..., qual è il modello più accurato, in termini di fedeltà dei risultati, che presenta il minor costo computazionale? Il problema è stato affrontato attraverso due differenti approcci, che hanno portato alla creazione della "Best Plate Theory Curve", attraverso la quale è possibile, per un dato problema, identificare il modello più idoneo in termini di fedeltà dei risultati e di costo computazionale. Nella seconda parte del lavoro viene presentato un modello mono-dimensionale multi-campo avanzato in grado di cogliere l’inserzione di elementi piezo-elettrici in ali in materiale composito. Questo elemento è stato viluppato partendo dalla Carrera Unified Formulation e dalle equazioni costitutive elettromeccaniche. Sono state effettuate poi delle validazioni attraverso confronti con la bibliografia.

Nonlinear time history analysis is increasingly being used in the design of tall steel structures, but member sizes still must be determined by a designer before an analysis can be performed. Often the distribution of story strength is still based on an assumed first mode response as determined from the Equivalent Lateral Force (ELF) procedure. For tall buckling restrained braced frames (BRBFs), two questions remain unanswered: what brace distribution will minimize total brace area, while satisfying story drift and ductility limits, and is the ELF procedure an effective approximation of that distribution? In order to investigate these issues, an optimization algorithm was incorporated into the OpenSees dynamic analysis platform. The resulting program uses a genetic algorithm to determine optimum designs that satisfy prescribed drift/ductility limits during nonlinear time history analyses. The computer program was used to investigate the optimized distribution of brace strength in BRBFs with different heights. The results of the study provide insight into efficient design of tall buildings in high seismic areas and evaluate the effectiveness of the ELF procedure.

<p dir="ltr">There is a growing demand for lower-cost, lighter-weight, and more compact electric ma-<br>chines used for vehicle propulsion. In this research, a dual-rotor inert-core machine (ICM) is<br>considered to meet this demand. In the ICM, permanent-magnet-based Halbach arrays are<br>placed on inner and outer rotating structures. This enables one to eliminate magnetic steels<br>used in the stator and rotor of traditional electric machinery. In addition, a stator structure<br>that leverages a thermal plastic is proposed that facilitates straightforward active cooling<br>of phase windings, which greatly increases current density. To support the multi-objective<br>design of the ICM, a multi-physics toolbox has been developed. Within the toolbox, electro-<br>magnetic performance is predicted using a method-of-moments-based field solver. Thermal<br>performance is assessed using a thermal equivalent circuit that includes conductive heat<br>transfer from stator windings to the surrounding environment as well as convective heat<br>transfer to moving fluids. The structural integrity of the stator is assessed using analytical<br>expressions to predict stress from material properties, geometry, and applied external forces.<br>Calculated loss of proposed designs includes those of the stator windings as well as those<br>required for active cooling. Several optimization studies have been conducted to evaluate the<br>performance of the ICM under an expected electric vehicle driving cycle. From the studies,<br>Pareto-optimal fronts are obtained and used to explore the impact of alternative cooling<br>strategies on volumetric power density.</p>

Dans un contexte de mobilité ubiquitaire, des différents objets sont capables d'interagir avec les utilisateurs pour leurs fournir des services innovants et les aider à optimiser leurs plans de déplacement. En effet, le nombre des utilisateurs est ainsi que le nombre de fournisseurs de service demandés par ces utilisateurs sont en pleine augmentation. Cette croissance implique un aspect de concurrence et nécessite des choix optimisés. Dans ce cadre, l'objectif de cette thèse est de concevoir et optimiser un système d'aide au déplacement qui couvre non seulement les services de déplacement quotidien mais aussi les services touristiques, culturels et bien d'autres. Les travaux de recherche présentés dans ce manuscrit proposent la mise en place d'une Plateforme de Recherche et de composition des Services d'Aide à la Mobilité (PRoSAM) afin d'optimiser les tâches de recherche, composition et distribution des Informations de Mobilité Avancée (IMA). L'aspect dynamique et distribué du problème, nous a conduit à adopter une modélisation orientée agent afin de s'adapter aux conditions d'un environnement ubiquitaire. Grâce à une approche de changement dynamique des rôles des agents utilisateurs et un protocole de négociation innovant, les clients sont capables d'échanger les services d'une manière autonome et d’établir des accords totaux ou partiels en fonction de l'offre et de la demande. Finalement, les résultats de simulation présentés dans cette thèse démontrent l'efficacité des approches adoptées / In a context of ubiquitous mobility, different objects are able to interact with users to provide them with innovative services and help them optimize their travel plans. Indeed, the number of users and the number of service providers requested by these users are actively growing. This growth involves an aspect of competition and requires optimized choices. In this context, the aim of this thesis is to design and optimize a mobility aid system that covers not only transport services but also tourist services, cultural services and many others. The research presented in this thesis proposes the establishment of a Plateforme de Recherche et de composition des Services d'Aide à la Mobilité (PRoSAM) to optimize research, composition and distribution tasks of advanced mobility information. The dynamic and distributed aspects of the problem have led us to adopt an agent-oriented modeling approach to cope with the conditions of a ubiquitous environment. Thanks to a dynamic role switching strategy of user agents and an innovative negotiation protocol, customers are able to exchange services autonomously and to establish full or partial agreements in order to optimize communications over the network. Finally, the simulation results presented in this thesis demonstrate the effectiveness of the proposed approaches

Rising concerns related to the effects of traffic congestion have led to the search for alternative transportation solutions. Advances in battery technology have resulted in an increase of electric vehicles (EVs), which serve to reduce the impact of many of the negative consequences of congestion, including pollution and the cost of wasted fuel. Furthermore, the energy-efficiency and quiet operation of electric motors have made feasible concepts such as Urban Air Mobility (UAM), in which electric aircraft transport passengers in dense urban areas prone to severe traffic slowdowns. Electrified transportation may be the solution needed to combat urban gridlock, but many logistical questions related to the design and operation of the resultant transportation networks remain to be answered. This research begins by examining the near-term effects of EV charging networks. Stationary plug-in methods have been the traditional approach to recharge electric ground vehicles; however, dynamic charging technologies that can charge vehicles while they are in motion have recently been introduced that have the potential to eliminate the inconvenience of long charging wait times and the high cost of large batteries. Using an agent-based model verified with traffic data, different network designs incorporating these dynamic chargers are evaluated based on the predicted benefit to EV drivers. A genetic optimization is designed to optimally locate the chargers. Heavily-used highways are found to be much more effective than arterial roads as locations for these chargers, even when installation cost is taken into consideration. This work also explores the potential long-term effects of electrified transportation on urban congestion by examining the implementation of a UAM system. Interdependencies between potential electric air vehicle ranges and speeds are explored in conjunction with desired network structure and size in three different regions of the United States. A method is developed to take all these considerations into account, thus allowing for the creation of a network optimized for UAM operations when vehicle or topological constraints are present. Because the optimization problem is NP-hard, five heuristic algorithms are developed to find potential solutions with acceptable computation times, and are found to be within 10% of the optimal value for the test cases explored. The results from this exploration are used in a second agent-based transportation model that analyzes operational parameters associated with UAM networks, such as service strategy and dispatch frequency, in addition to the considerations associated with network design. General trends between the effectiveness of UAM networks and the various factors explored are identified and presented.