RESUMO Simulações de aeroacústica computacional demandam uma quantidade considerável de tempo, o que torna complicada a realização de estudos paramétricos. O presente trabalho propõe uma metodologia viável para otimização aeroacústica. Através da análise numérica utilizando dinâmica dos fluidos computacional, foi estudada a aplicação de uma placa separadora desacoplada como método de controle passivo da esteira turbulenta de um cilindro e avaliou-se a irradiação de ruído causado pela interação do escoamento com ambos os corpos, empregando ferramentas de aeroacústica computacional baseadas no método de Ffowcs-Williams e Hawkings. Algumas abordagens distintas de metodologias de otimização de projeto foram aplicadas neste problema, com o objetivo de chegar a uma configuração otimizada que permita a redução do nível sonoro ao longe. Assim, utilizando uma ferramenta de otimização multidisciplinar, pode-se avaliar a capacidade de modelos heurísticos e a grande vantagem do emprego de algoritmos baseados em método de superfície de resposta quando aplicados em um problema não linear, pois requerem a avaliação de um menor número de alternativas para se obter um ponto ótimo. Além disso, foi possível identificar e agrupar os resultados em 5 clusters baseados em seus parâmetros geométricos, nível de pressão sonora global e o valor quadrático médio do coeficiente de arrasto, confirmando a eficiência da aplicação de placas separadoras longas desacopladas posicionadas próximas ao cilindro na estabilização da esteira turbulenta, enquanto que o posicionamento de placas acima de um espaçamento crítico aumentou o nível de pressão acústica irradiado devido à formação de vórtices no espaço entre o cilindro e a placa separadora. / Computational aeroacoustics simulations require a considerable amount of time, which makes the comparison of a large number of different geometric designs a difficult task. The goal of the present study is to provide a suitable methodology for aeroacustic optimization. By means of numerical analyses using computational fluid dynamics tools, the application of a detached splitter plate as a passive control method for the turbulent wake of a circular cylinder was investigated. The irradiation of noise caused by the interaction between the flow and both bodies was evaluated using computational aeroacoustics tools based on the Ffowcs-Williams and Hawkings method. Various design optimization methodologies were applied to this flow in order to achieve a possible optimal configuration, i.e., one which is capable of reducing the far field noise level without increasing the aerodynamic forces. Using a multidisciplinary optimization tool, it was possible to evaluate the behavior of heuristic optimization algorithms and the major advantage of algorithms based on response surface methods when applied to a nonlinear aeroacoustics problem, since they require a smaller number of calculated designs to reach the optimal configuration. In addition, it was possible to identify and group the outcomes into 5 clusters based on their geometric parameters, overall sound pressure level and drag coefficient, confirming the efficiency of the application of long detached splitter plates placed next to the cylinder in stabilizing the turbulent wake, whereas the positioning of splitter plates at a distance larger than a critical gap increased the overall sound pressure level radiated due to the formation of vortices in the gap.

Neste trabalho foram desenvolvidas duas técnicas de sintonia para controladores preditivos por modelo. Ambas visam minimizar a soma do erro quadrático entre respostas do sistema em malha fechada e trajetórias de referência pré-definidas; a primeira resolve um problema de otimização lexicográfica enquanto a segunda resolve um problema de otimização de compromisso. As vantagens dos métodos apresentados são: maior automatização, definição de objetivos de sintonia intuitiva que considera especificações na dinâmica do processo, uma métrica no domínio do tempo e é capaz de incluir o conhecimento do engenheiro de controle em uma técnica de sintonia confiável. Um estudo de caso no sistema de craqueamento catalítico ilustrou a flexibilidade de definição dos objetivos da técnica lexicográfica. Um estudo de caso sobre uma coluna de fracionadora de óleo pesado em malha fechada com um controlador preditivo por modelo comparou ambas as estratégias de sintonia desenvolvidas aqui e pode-se concluir que a técnica lexicográfica dá prioridade aos objetivos importantes enquanto a técnica de compromisso calcula uma solução média, com respeito aos objetivos. A técnica de compromisso foi comparada a um método de sintonia da literatura quanto a aplicação em um controlador preditivo de horizonte infinito com targets para as entradas e controle por faixas das saídas com uma coluna de destilação. Observou-se que a técnica desenvolvida aqui é computacionalmente mais rápida e não requer a escolha de uma solução não-dominada dentre um conjunto de soluções de Pareto. Aplicações reais de controle preditivo são severamente afetadas por incerteza de modelo. Estendeu-se as técnicas desenvolvidas aqui para considerar o caso de incerteza multi-planta, calculando parâmetros de sintonia robustos para controladores nominais, visando tratar o compromisso entre performance e estabilidade e robustez da malha fechada. Um controlador preditivo de horizonte infinito foi sintonizado de forma robusta e comparado com um controlador preditivo robusto em malha fechada com um modelo de separadora C3/C4. Observou-se que este consegue controlar melhor o processo, entretanto, tem um tempo de computação duas ordens de grandeza maior que o controlador nominal, em operação on-line. / Two multi-objective optimization based tuning techniques for Model Predictive Control (MPC) were developed. Both take into account the sum of the squared errors between closed-loop trajectories and reference responses based on pre-defined goals as tuning objectives; one solves a lexicographic optimization to obtain an optimum set of tuning parameters (LTT), whereas the other solves a compromise optimization problem (CTT). The main advantages are an automated framework, and straightforward goal definition, which are capable of taking into account a specification on the process dynamics, a time-domain metrics, and of embedding the control engineers knowledge into a reliable approach. A fluid catalytic cracking tuning case study unveiled the goal definition flexibility of the LTT, with respect to output tracking and variable coupling. A heavy oil fractionator in closed-loop with a MPC case study compared both tuning techniques developed here, and it was observed that the LTT in fact prioritizes the main objectives, whereas the CTT yields an average solution, in terms of the tuning objectives. The CTT was compared to another multi-objective tuning technique from the literature, in the tuning of a MPC with input targets and output zone control in closed-loop with a crude distillation unit model. The simulation results showed that the CTT allows for faster results, regarding the computational time to compute the tuning parameters and there is no need of a posteriori decisions to select the best non-dominated solution. Real MPC applications are strongly hindered by model uncertainty. This limitation was addressed by the extension of the tuning techniques to account for multi-plant model uncertainty, thus obtaining optimum robustly tuned parameters for nominal controllers, addressing the trade-off between robustness and performance. A robustly tuned Infinite Horizon MPC (IHMPC) was compared to a Robust IHMPC, in closed-loop with a C3/C4 splitter system model. It was observed in a simulation that even though the latter yields better output responses, it is two orders of magnitude slower than the former in online operation.

Orientador: Fernando José Von Zuben / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-26T06:49:07Z (GMT). No. of bitstreams: 1 Azevedo_CarlosRenatoBelo_D.pdf: 3449858 bytes, checksum: 7a1811aa772f1ae996e8851c60627b7c (MD5) Previous issue date: 2012 / Resumo: A presença de incerteza em resultados futuros pode levar a indecisões em processos de escolha, especialmente ao elicitar as importâncias relativas de múltiplos critérios de decisão e de desempenhos de curto vs. longo prazo. Algumas decisões, no entanto, devem ser tomadas sob informação incompleta, o que pode resultar em ações precipitadas com consequências imprevisíveis. Quando uma solução deve ser selecionada sob vários pontos de vista conflitantes para operar em ambientes ruidosos e variantes no tempo, implementar alternativas provisórias flexíveis pode ser fundamental para contornar a falta de informação completa, mantendo opções futuras em aberto. A engenharia antecipatória pode então ser considerada como a estratégia de conceber soluções flexíveis as quais permitem aos tomadores de decisão responder de forma robusta a cenários imprevisíveis. Essa estratégia pode, assim, mitigar os riscos de, sem intenção, se comprometer fortemente a alternativas incertas, ao mesmo tempo em que aumenta a adaptabilidade às mudanças futuras. Nesta tese, os papéis da antecipação e da flexibilidade na automação de processos de tomada de decisão sequencial com múltiplos critérios sob incerteza é investigado. O dilema de atribuir importâncias relativas aos critérios de decisão e a recompensas imediatas sob informação incompleta é então tratado pela antecipação autônoma de decisões flexíveis capazes de preservar ao máximo a diversidade de escolhas futuras. Uma metodologia de aprendizagem antecipatória on-line é então proposta para melhorar a variedade e qualidade dos conjuntos futuros de soluções de trade-off. Esse objetivo é alcançado por meio da previsão de conjuntos de máximo hipervolume esperado, para a qual as capacidades de antecipação de metaheurísticas multi-objetivo são incrementadas com rastreamento bayesiano em ambos os espaços de busca e dos objetivos. A metodologia foi aplicada para a obtenção de decisões de investimento, as quais levaram a melhoras significativas do hipervolume futuro de conjuntos de carteiras financeiras de trade-off avaliadas com dados de ações fora da amostra de treino, quando comparada a uma estratégia míope. Além disso, a tomada de decisões flexíveis para o rebalanceamento de carteiras foi confirmada como uma estratégia significativamente melhor do que a de escolher aleatoriamente uma decisão de investimento a partir da fronteira estocástica eficiente evoluída, em todos os mercados artificiais e reais testados. Finalmente, os resultados sugerem que a antecipação de opções flexíveis levou a composições de carteiras que se mostraram significativamente correlacionadas com as melhorias observadas no hipervolume futuro esperado, avaliado com dados fora das amostras de treino / Abstract: The presence of uncertainty in future outcomes can lead to indecision in choice processes, especially when eliciting the relative importances of multiple decision criteria and of long-term vs. near-term performance. Some decisions, however, must be taken under incomplete information, what may result in precipitated actions with unforeseen consequences. When a solution must be selected under multiple conflicting views for operating in time-varying and noisy environments, implementing flexible provisional alternatives can be critical to circumvent the lack of complete information by keeping future options open. Anticipatory engineering can be then regarded as the strategy of designing flexible solutions that enable decision makers to respond robustly to unpredictable scenarios. This strategy can thus mitigate the risks of strong unintended commitments to uncertain alternatives, while increasing adaptability to future changes. In this thesis, the roles of anticipation and of flexibility on automating sequential multiple criteria decision-making processes under uncertainty are investigated. The dilemma of assigning relative importances to decision criteria and to immediate rewards under incomplete information is then handled by autonomously anticipating flexible decisions predicted to maximally preserve diversity of future choices. An online anticipatory learning methodology is then proposed for improving the range and quality of future trade-off solution sets. This goal is achieved by predicting maximal expected hypervolume sets, for which the anticipation capabilities of multi-objective metaheuristics are augmented with Bayesian tracking in both the objective and search spaces. The methodology has been applied for obtaining investment decisions that are shown to significantly improve the future hypervolume of trade-off financial portfolios for out-of-sample stock data, when compared to a myopic strategy. Moreover, implementing flexible portfolio rebalancing decisions was confirmed as a significantly better strategy than to randomly choosing an investment decision from the evolved stochastic efficient frontier in all tested artificial and real-world markets. Finally, the results suggest that anticipating flexible choices has lead to portfolio compositions that are significantly correlated with the observed improvements in out-of-sample future expected hypervolume / Doutorado / Engenharia de Computação / Doutor em Engenharia Elétrica

As a result of the increase in accessibility of computational resources and the increase in the power of the computers during the last two decades, designers are able to create computer models to simulate the behavior of a complex products. To address global competitiveness, companies are forced to optimize their designs and products. Optimizing the design needs several runs of computationally expensive simulation models. Therefore, using metamodels as an efficient and sufficiently accurate approximate of the simulation model is necessary. Radial basis functions (RBF) is one of the several metamodeling methods that can be found in the literature. The established approach is to add a bias to RBF in order to obtain a robust performance. The a posteriori bias is considered to be unknown at the beginning and it is defined by imposing extra orthogonality constraints. In this thesis, a new approach in constructing RBF with the bias to be set a priori by using the normal equation is proposed. The performance of the suggested approach is compared to the classic RBF with a posteriori bias. Another comprehensive comparison study by including several modeling criteria, such as problem dimension, sampling technique and size of samples is conducted. The studies demonstrate that the suggested approach with a priori bias is in general as good as the performance of RBF with a posteriori bias. Using the a priori RBF, it is clear that the global response is modeled with the bias and that the details are captured with radial basis functions. Multi-objective optimization and the approaches used in solving such problems are briefly described in this thesis. One of the methods that proved to be efficient in solving multi-objective optimization problems (MOOP) is the strength Pareto evolutionary algorithm (SPEA2). Multi-objective optimization of a disc brake system of a heavy truck by using SPEA2 and RBF with a priori bias is performed. As a result, the possibility to reduce the weight of the system without extensive compromise in other objectives is found. Multi-objective optimization of material model parameters of an adhesive layer with the aim of improving the results of a previous study is implemented. The result of the original study is improved and a clear insight into the nature of the problem is revealed.

Dans le processus de conception des véhicules, la vibration et le bruit sont des sujets d’étude très importants. En effet, les vibrations sont susceptibles d’affecter le comportement dynamique des structures et le bruit dégrade le confort acoustique des passagers. L’objectif principal de la thèse est de proposer un ensemble de méthodes pour l’optimisation du comportement dynamique des systèmes complexes afin de réduire les vibrations des structures et le bruit dans l’habitacle. À cet effet, on s’intéresse à des stratégies de contrôle des interfaces, comme le collage de couches viscoélastiques sur les zones les plus déformées, ou l’introduction de dispositifs frottants calibrés pour ajouter de l’amortissement à certaines fréquences de résonance. Les structures assemblées résultantes sont étudiées numériquement par une méthode de synthèse modale généralisée. La méthode de synthèse modale proposée contient plusieurs niveaux de condensation. Le premier concerne les degrés de libertés (DDL) internes de chaque sous-structure. La deuxième condensation s’effectue sur les modes de branches, de sorte à réduire le nombre de DDL aux interfaces entre les sous-structures. Pour les systèmes couplés fluide/structure, une troisième condensation portant sur les DDL du fluide est proposée. Suite à ces condensations, la dimension du système est fortement réduite. Cette méthode permet alors d’obtenir aussi bien la réponse forcée de la structure que les fluctuations du champ de pression dans le fluide. Les chemins de transmission acoustiques et vibratoires peuvent également être déduits des contributions modales intermédiaires. On montre que ces paramètres modaux peuvent être utilisés comme fonctions objectif pour une démarche d’optimisation des interfaces. Le front de Pareto des conceptions optimales est obtenu avec un algorithme génétique multi-objectif élitiste, appliqué à une approximation par krigeage de la fonction objectif. Cette approche modale est étendue à l’étude de systèmes non-linéaires. L’hypothèse fondamentale est que les modes non-linéaires sont faiblement couplés. Les paramètres modaux non-linéaires (fréquences propres, amortissements...), dépendent des amplitudes modales. L’idée est alors de calculer des modes normaux non-linéaires en fonction de leur amplitude et de superposer leurs réponses pour obtenir celle de la structure. La méthode est appliquée à des systèmes incorporant des non-linéarités de type Duffing et de frottement sec. Le cas particulier du frottement sec est considéré à travers un modèle de Masing généralisé. Deux approches modales sont développées : l’une basée sur les modes complexes, et l’autre basée sur les modes réels. L’utilisation de modes complexes ou réels dans la synthèse modale conduit à des termes d’amortissement par frottement différents. On montre que la synthèse modale non-linéaire combinée au modèle de Masing généralisé aboutit à une méthode numérique simple, rapide et efficace pour décrire le comportement non-linéaire de structures soumise à du frottement sec. / Noise and vibration are important topics in the automotive industry for several reasons, including passenger comfort and structural integrity. The main objective of this thesis is to propose a series of appropriate methods to optimize structural system characteristics, so that the vibration and noise can be reduced. To achieve this goal, interface control strategies are employed, including bonding viscoelastic layers onto the most heavily deformed zones and introducing frictional damping devices calibrated on certain resonance frequencies. Such built-up structural systems are numerically investigated via a generalized modal synthesis approach that incorporates several groups of modes. The employed modal synthesis approach consists of several levels of condensation. The first one is on the internal degrees of freedoms (DOFs) of each substructure, and the second condensation is on the branch modes so as to reduce the boundary DOFs among substructures. For coupled fluid-structural systems, a third condensation on the fluid DOFs is suggested. With these condensation techniques, the system dimension can be significantly reduced. The method allows us to obtain the forced response of the structures as well as the pressure variation of the fluids. Additionally, modal parameters characterizing vibration and noise transmission paths can be deduced as mid-stage results. We show that these modal parameters can be used as optimization objective during the interface configuration design. The Pareto front of the optimal design is achieved by employing Kriging approximations followed with an elitist multi-objective genetic algorithm. Another advantage of the modal approach is that a modal overview on the system characteristics is provided by analyzing the natural frequencies, modal damping ratios and the aforementioned modal parameters. The modal synthesis approach is further extended to study nonlinear systems. The basic assumption is that the nonlinear modes are weakly coupled. Nonlinear modal parameters, such as modal frequency and modal damping ratio, contain the essential nonlinear information and depend on modal amplitude. The main idea is to compute nonlinear normal modes according to their modal amplitude and superimpose the response of several nonlinear modes to obtain the overall forced response. The method is applied to systems involving Duffing and dry friction nonlinearities. In the case of dry friction, a generalized Masing model is considered to capture the dry friction nature. Both complex modes and real modes are used in the modal synthesis, leading to different frictional damping terms. We show that the nonlinear modal synthesis combined with the generalized Masing model yields a simple, fast and efficient numerical method to describe nonlinear performance of structures with dry friction.

L'hadronthérapie est une modalité d'irradiation récente particulièrement attractive. Les ions, par leur profil caractéristique de dépôt de dose dans la matière et leur efficacité biologique accrue, sont des particules parfaitement adaptées pour le traitement du cancer. C’est une modalité émergente et les travaux de recherche et de développement qui en font l'objet se poursuivent de manière soutenue. Cependant, il n'existe à ce jour aucun outil permettant de quantifier pour le patient le gain clinique associé aux améliorations proposées, comme l’apport d’une gantry par exemple. Nous proposons dans ce travail de concevoir un module de planification inverse du traitement pour un but prospectif. Détachée des contraintes usuelles de précision et de temps de calcul, notre méthode d'optimisation se base sur un algorithme génétique afin d'approcher d'une solution globale vis-à-vis d'un grand nombre de paramètres balistiques (champs d’irradiation libres) et en associant les diverses régions d'intérêt dosimétrique. La stratégie d'optimisation retenue est progressivement complexifiée afin de prendre en compte de manière efficace les différents enjeux de la planification. La robustesse du plan vis-à-vis des incertitudes inhérentes au traitement, primordiale en hadronthérapie, est évaluée. Dans toutes les situations testées, il apparaît que l'inclusion dans l'optimisation de paramètres habituellement fixés manuellement permet une amélioration de la qualité de traitement. Nous proposons au terme de cette étude un outil prospectif d'optimisation au réglage simple et capable de mener des études comparées sur la pertinence de nouvelles modalités d’irradiation. / Hadrontherapy is a recent and particularly attractive modality. Characterized by a specific dose deposition profile in matter and by a high biological effectiveness, ions are found to be very well-suited for cancer treatment. As an emergent modality, the research in hadrontherapy is extremely active and promises many improvements for the future. However, there is no tool to date to quantify the clinical benefit for the patient related to the proposed improvements, the use of a gantry for example. In this work, we propose to use the treatment planning system, usually dedicated to clinical practice, in a prospective purpose. Suppressing the classical constraints of precision and time, our optimization method is based on a genetic algorithm designed to approach a global solution including a high number of balistic parameters (free irradiation fields) for all regions of dosimetric interest. The optimization strategy is progressively complicated in order to efficientely take into account the main issues of the inverse planning problem. The robustness of plans towards the uncertainties related to the application of the treatment, essential in hadrontherapy, is evaluated. The results show that the inclusion, in the optimization, of parameters usually fixed by the human planner leads systematically to an improved treatment quality. The final product of this work is a prospective optimization tool characterized by an easy set-up system and the ability to perform comparative studies on the relevance of new irradiation modalities.

Le but de ce travail est de proposer une démarche structurée pour l'étude des liaisons lubrifiées bielle/vilebrequin. Bien que cette source de déperdition énergétique ne soit pas primordiale, il n'en est pas moins vrai que la maîtriser est un enjeu pour les motoristes. Les normes anti-pollution imposées aux constructeurs automobiles étant de plus en plus strictes, un effort doit être réalisé sur l’ensemble des composants afin de réduire de façon significative les pertes d’énergie dans les moteurs. La tendanceactuelle est à la diminution de la viscosité du lubrifiant, toutefois ce n'est pas sans conséquence sur l'intégrité des liaisons.La démarche proposée, qui s'appuie sur un logiciel TEHD consiste à utiliser la méthode des plans d’expériences afin de créer des métamodèles de la puissance dissipée et de la sévérité du cas traité. Une optimisation multi-objectifs (déperdition et sévérité) par algorithme évolutionnaire et utilisant les métamodèles est ensuite menée afin de déterminer les combinaisons de facteurs offrant les meilleurs compromis par rapport à ces deux objectifs. La méthodologie proposée s'avère pertinente puisque des modèles analytiques simples se substituent très avantageusement aux lourds calculs TEHD.Afin de valider le logiciel TEHD sur l'évaluation des puissances dissipées, une comparaison numérique/expérimentale est conduite sur un banc d’essais permettant de reproduire les conditions réelles de fonctionnement des bielles dans les moteurs. / The aim of this work is to propose a structured approach of the study of the lubricated connection between the conrod and the crankshaft. Although this source of energy loss is not the most significant one, improving this contact remains a challenging task for engine manufacturers. The anti-pollution standards imposed on automakers are increasingly stringent, an effort must be made on all components in order to significantly reduce energy loss in the engines. The current trend is to decrease the viscosity of thelubricant, however, it is not without consequences for the connections health. The proposed approach, which relies on a TEHD software is to use the design of experiments method to create surrogates of both power loss and severity for the treated case. A multi-objective optimization (power loss and severity) by evolutionary algorithm and using the surrogates, is then performed to determine the right combination of factors that provides the best compromise with respect to these two objectives. The proposed methodology is relevant because simple analytical models replace very favorably time consuming TEHD calculations.To validate the evaluation of power loss calculated with the TEHD software, a numerical/experimental comparison is performed on a test bench which reproduces the real operating conditions of the connecting rods.

Le domaine de la traction électrique a suscité un très grand intérêt dans les dernières années. La conception optimale de l'ensemble moteur électrique de traction – onduleur doit prendre en compte une variété de critères et contraintes. Étant donnée la liaison entre la géométrie du moteur et la stratégie de commande de l'onduleur, l'optimisation de l'ensemble de traction doit prendre en considération, en même temps, les deux composants.L'objectif de la thèse est la conception d'un outil d'optimisation appliqué à un système de traction électrique légère qu'emploie un moteur à réluctance variable alimenté (MRVCE) par un onduleur triphasé en pont complet. Le MRVCE est modélisé en utilisant la technique par réseau de perméances. En même temps, la technique de commande électronique peut être facilement intégrée dans le modèle pour effectuer l'analyse dynamique du fonctionnement du moteur. L'outil d'optimisation réalisé utilise l'algorithme par essaim de particules, modifié pour résoudre des problèmes multi-objectif. Les objectifs sont liés à la qualité des caractéristiques de fonctionnement du moteur, en temps que les variables d'optimisation concernent la géométrie du moteur aussi que la technique de commande. Les performances de l'algorithme sont comparées avec ceux de l'algorithme génétique (NSGA-II) et d'une implémentation classique de l'algorithme par essaim de particules multi-objectif.Finalement, un prototype de moteur à réluctance variable est construit et le fonctionnement du MRVCE alimenté depuis l'onduleur triphasé en pont complet est implémenté et les outils de modélisation et d'optimisation sont validés / The interest for the electric traction applications has been growing in the last few years. The optimal design of the electric motor and of the inverter that powers it needs to consider a long list of restrictions and criteria. Because of the fact that the geometry of the motor and the switching strategy are closely linked, the optimization of the traction solution needs to consider both, at the same time.The objective of this thesis is the development of an optimization tool applied for the optimization of an electric traction solution that uses the switched reluctance motor (SRM) fed from a three phase full bridge inverter. The SRM is modeled using Permeance Network Analysis (PNA). The switching technique can be easily integrated in the model, which gives the possibility to run a dynamic analysis. The optimization tool created uses the Particle Swarm Optimization (PSO) algorithm, modified for multi-objective problems. The algorithms performances are compared with those of the Genetic Algorithm, using the NSGA-II multi-objective technique and with a classic version of multiple objective particle swarm optimizer (MOPSO).Finally, a SRM prototype is constructed and the drive solution using a full-bridge three phase inverter is implemented. The modeling and optimization tools are thus experimentally validated

Cette thèse traite de l’optimisation multiobjectif de fonctions coûteuses, aboutissant à laconstruction d’un front de Pareto représentant l’ensemble des compromis optimaux. En conception automobile, le budget d’évaluations est fortement limité par les temps de simulation numérique des phénomènes physiques considérés. Dans ce contexte, il est courant d’avoir recours à des « métamodèles » (ou modèles de modèles) des simulateurs numériques, en se basant notamment sur des processus gaussiens. Ils permettent d’ajouter séquentiellement des observations en conciliant recherche locale et exploration. En complément des critères d’optimisation existants tels que des versions multiobjectifs du critère d’amélioration espérée, nous proposons d’estimer la position de l’ensemble du front de Pareto avec une quantification de l’incertitude associée, à partir de simulations conditionnelles de processus gaussiens. Une deuxième contribution reprend ce problème à partir de copules. Pour pouvoir traiter le cas d’un grand nombre de variables d’entrées, nous nous basons sur l’algorithme REMBO. Par un tirage aléatoire directionnel, défini par une matrice, il permet de trouver un optimum rapidement lorsque seules quelques variables sont réellement influentes (mais inconnues). Plusieurs améliorations sont proposées, elles comprennent un noyau de covariance dédié, une sélection du domaine de petite dimension et des directions aléatoires mais aussi l’extension au casmultiobjectif. Enfin, un cas d’application industriel en crash a permis d’obtenir des gainssignificatifs en performance et en nombre de calculs requis, ainsi que de tester le package R GPareto développé dans le cadre de cette thèse. / This dissertation deals with optimizing expensive or time-consuming black-box functionsto obtain the set of all optimal compromise solutions, i.e. the Pareto front. In automotivedesign, the evaluation budget is severely limited by numerical simulation times of the considered physical phenomena. In this context, it is common to resort to “metamodels” (models of models) of the numerical simulators, especially using Gaussian processes. They enable adding sequentially new observations while balancing local search and exploration. Complementing existing multi-objective Expected Improvement criteria, we propose to estimate the position of the whole Pareto front along with a quantification of the associated uncertainty, from conditional simulations of Gaussian processes. A second contribution addresses this problem from a different angle, using copulas to model the multi-variate cumulative distribution function. To cope with a possibly high number of variables, we adopt the REMBO algorithm. From a randomly selected direction, defined by a matrix, it allows a fast optimization when only a few number of variables are actually influential, but unknown. Several improvements are proposed, such as a dedicated covariance kernel, a selection procedure for the low dimensional domain and of the random directions, as well as an extension to the multi-objective setup. Finally, an industrial application in car crash-worthiness demonstrates significant benefits in terms of performance and number of simulations required. It has also been used to test the R package GPareto developed during this thesis.

Les stations d’épuration se tournent actuellement vers des installations de récupération d'énergie et des nutriments. Dans le même temps, elles sont soumises à une réglementation de plus en plus stricte en ce qui concerne l'environnement et la santé humaine. Face au défi ambitieux de réduire les coûts d'exploitation et les impacts environnementaux tout en garantissant la robustesse du procédé, il est nécessaire de développer des outils capables de fournir une évaluation intégrée du procédé. L’objectif de ce travail est de développer une plateforme réaliste et prédictive contenant trois aspects: la modélisation rigoureuse et dynamique de l’ensemble de la station d’épuration; l'analyse de cycle de vie aux frontières étendues pour l'évaluation des scénarios et enfin un outil d'optimisation multi-objectif efficace. La plateforme développée pour l'évaluation environnementale couplée à la modélisation dynamique a d'abord été appliquée à plusieurs cas d’étude. Ainsi des résultats de performance et d’impacts environnementaux ont été obtenus pour la séparation de l’urine à la source, la décantation primaire avancée et le traitement de l'urine par nitritation/ oxydation anaérobie de l’ammonium, et d’autres filières. Compte tenu des importants avantages de la séparation de l’urine établis par les résultats précédents, un générateur d’influents phénoménologique, flexible et dynamique a été adapté afin de fournir des données dynamiques réalistes concernant les flux d'urine et des eaux usées dans les différents scénarios de rétention d'urine. Enfin, comme la combinaison complexe de processus biologiques, chimiques et physiques conduit à un problème lourd en calcul, une étude de faisabilité (temps de calcul et fiabilité) a été réalisée sur l'optimisation multi-objectif. L'obtention d'un ensemble de solutions qui évite toute discrimination préalable entre les coûts, l'environnement et les performances ont permis la discussion des enjeux impliqués. Finalement, la plateforme complète a été appliquée à plusieurs cas d’étude et clarifie les aspects opérationnels des options plus durables en matière de gestion et de traitement des eaux usées. / Wastewater treatment plants are moving towards energy and nutrients recovery facilities. Simultaneously, they are submitted to stricter regulation with respect to environment and human health. Facing the great challenge of reducing operational costs along with the reduction of environmental impacts and the guaranty of plants robustness, tools might be developed in order to provide an integrated assessment. The goal of this work is to develop a reliable and predictive framework containing rigorous dynamic wide-plant modelling, extended boundaries life cycle assessment for scenarios evaluation and an efficient multi-objective optimization tool. The developed framework for environmental evaluation coupled to dynamic modelling was initially applied to several case studies including urine source separation, enhanced primary clarification and urine treatment by nitritation/ anaerobic ammonium oxidation, offering both performance results and environmental hotspots. Given the important benefits of the urine source separation provided by the previous results, a flexible and dynamic phenomenological influent generator was adapted in order to provide realistic dynamic data concerning urine and wastewater streams in different urine retention scenarios. Finally, as the complex combination of biological, chemical and physical processes leads to a computational expensive problem, a feasibility study (computational time and reliability) on the multi-objective optimization was conducted. Obtaining a set of solutions that avoids any prior discrimination among costs, environment and performance allowed thus the discussion of the involved trade-offs. Finally, the complete framework was applied to several case studies lightening on operational aspects of more sustainable options on wastewater management and treatment.