In modern day engineering, the designer has become more and more dependent on computer simulation. Oftentimes, computational cost and convergence accuracy accompany these simulations to reach global solutions for engineering design problems causes traditional optimization techniques to perform poorly. To overcome these issues nontraditional optimization algorithms based region elimination and space exploration are introduced. Approximation models, which are also known as metamodels or surrogate models, are used to explore and give more information about the design space that needs to be explored. Usually the approximation models are constructed in the promising regions where global solutions are expected to exist. The approximation models imitate the original expensive function, black-box function, and contribute towards getting comparably acceptable solutions with fewer resources and at low computation cost. The primary contributions of this dissertation are associated with the development of new methods for exploring the design space for large scale computer simulations. Primarily, the proposed design space exploration procedure uses a hierarchical partitioning method to help mitigate the curse of dimensionality often associated with the analysis of large scale systems. The research presented in this dissertation focuses on introducing new optimization algorithms based on metamodeling techniques that alleviate the burden of the computation cost associated with complex engineering design problems. Three new global optimization algorithms were introduced in this dissertation, Approximated Unimodal Region Elimination (AUMRE), Space Exploration and Unimodal Region Elimination (SEUMRE), and Mixed Surrogate Space Exploration (MSSE) for computation intensive and black-box engineering design optimization problems. In these algorithms, the design space was divided into many subspaces and the search was focused on the most promising regions to reach global solutions with the resources available and with less computation cost. Metamodeling techniques such as Response Surface Method (RSM), Radial Basis Function (RBF), and Kriging (KRG) are introduced and used in this work. RSM has been used because of its advantages such as being easy to construct, understand and implement. Also due to its smoothing capability, it allows quick convergence of noisy functions in the optimization. RBF has the advantage of smoothing data and interpolating them. KRG metamodels can provide accurate predictions of highly nonlinear or irregular behaviours. These features in metamodeling techniques have contributed largely towards obtaining comparably accurate global solutions besides reducing the computation cost and resources. Many multi-objective optimization algorithms, specifically those used for engineering problems and applications involve expensive fitness evaluations. In this dissertation, a new multi-objective global optimization algorithm for black-box functions is also introduced and tested on benchmark test problems and real life engineering applications. Finally, the new proposed global optimization algorithms were tested using benchmark global optimization test problems to reveal their pros and cons. A comparison with other well known and recently introduced global optimization algorithms were carried out to highlight the proposed methods’ advantages and strength points. In addition, a number of practical examples of global optimization in industrial designs were used and optimized to further test these new algorithms. These practical examples include the design optimization of automotive Magnetorheological Brake Design and the design optimization of two-mode hybrid powertrains for new hybrid vehicles. It is shown that the proposed optimization algorithms based on metamodeling techniques comparably provide global solutions with the added benefits of fewer function calls and the ability to efficiently visualize the design space. / Graduate

Orientador: Fernando Antonio Campos Gomide / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-05T20:02:48Z (GMT). No. of bitstreams: 1 MotaFilho_FranciscoOsvaldoMendes_M.pdf: 2700152 bytes, checksum: 3ab58e91f1a3839dae9d39e47d33ff50 (MD5) Previous issue date: 2005 / Resumo: Para obter uma solução satisfatória, algoritmos genéticos avaliam, em geral, um número grande de indivíduos durante o processo evolutivo. É comum, em aplicações práticas, encontrar funções de avaliação computacionalmente complexas e caras. Porém, nesses casos, o tempo é um fator determinante no desempenho de algoritmos genéticos. Dessa forma, os algoritmos genéticos devem encontrar soluções adequadas em curto intervalo de tempo. Uma alternativa promissora para contornar os custos computacionais referentes à função de avaliação considera o fato de que pode ser mais atrativo avaliar diretamente somente indivíduos selecionados e estimar os fitness dos restantes do que avaliar diretamente toda a população. Este trabalho propõe o uso de modelos de estimação de fitness em algoritmos genéticos. Especificamente, são sugeridos modelos de estimação baseados em agrupamento nebuloso supervisionado (Fuzzy C-Means) e não supervisionado (Aprendizagem Participativa). O objetivo é aproximar as funções de avaliação por meio de modelos de estimação de fitness, sem afetar significativamente a qualidade das soluções. Inicialmente, os modelos de estimação propostos são comparados e analisados experimentalmente com alternativas sugeri das por outros autores, utilizando, para isso, problemas de otimização considerados na literatura de algoritmos genéticos. A seguir, os modelos de estimação de fitness são aplicados em um problema real de engenharia, o planejamento de circulação de trens em ferrovias. Este é um caso típico onde o desempenho de cada planejamento exige um tempo significativo. A eficiência dos modelos propostos é verificada e comprovada experimentalmente comparando com os resultados, em instâncias mais simples, fornecidos por modelos de programação matemática e, em instâncias complexas, fornecidos pelo algoritmo genético clássico / Abstract: Genetic algorithms usually need a large number of fitness evaluations before a satisfying result can be obtained. In many real-world applications, fitness evaluation may be computationally complex and costly. In these cases, time is an essential subject in performance analysis of genetic algorithms. Therefore, genetic algorithms should provide good solutions in a short period of time. A promising approach to alleviate the computational cost of evaluations considers the fact that sometimes it is better to evaluate only selected individuals and estimate the fitness of the remaining individuals instead of evaluate a whole population. This work suggests the application of fitness estimation models in genetic algorithms. More specifically, it deals with estimation models based on supervised fuzzy clustering (Fuzzy C-Means) and unsupervised fuzzy clustering (Participatory Learning). The goal is to approximate the evaluation functions through the use of fitness estimation models, without significantly affect the quality of solutions. Initially, the fitness estimation models are compared and analyzed experimentally with other models already proposed in the literature. Their performance are evaluated using benchmark optimization problems found in the genetic algorithms literature. Next, the fitness estimation models are used to solve a real-world engineering problem, namely the train scheduling in a freight rail line. This is a typical case where the performance measure of each schedule demands a considerable amount of time. Once again, the performance of the fitness estimation models are evaluated experimentally, comparing their results with the results provided, for simple instances, by linear programming models and, for complex instances, by the classic genetic algorithm / Mestrado / Engenharia de Computação / Mestre em Engenharia Elétrica

Mit deterministischen Sicherheitsanalysen werden Auslegungsstörfälle bei Kernkraftwerken anhand von Rechenmodellen am Computer berechnet, um damit die Funktionalität der installierten Sicherheitssysteme eines jeden Kraftwerkes zu überprüfen. Allerdings sind bei solchen Störfalluntersuchungen stets Unsicherheiten vorhanden, die den zu ermittelnden Störfallablauf stark beeinflussen können. Beispielsweise können aufgrund technisch bedingter Fertigungstoleranzen schwankende Geometrie- und Materialdaten entstehen, die bei der Modellierung des Rechenmodells zu Unsicherheiten führen. Weitere Unsicherheiten können auf die physikalischen Modelle eines Störfall-analyseprogrammes zurückgeführt werden. Insbesondere haben die empirischen Beziehungen Unsicherheiten, da diese aus experimentellen Daten ermittelt wurden. In der vorliegenden Arbeit wurden daher die empirischen Beziehungen des Programmes TRACE analysiert und dessen Unsicherheiten quantifiziert. Mit der entwickelten „Dynamic Best-Estimate Safety Analysis“-Methode (DYBESA-Methode) lässt sich diese programmspezifische Unsicherheit bei der Störfalluntersuchung berücksichtigen. Es wurden 13 verschiedene „Correlation, Identification and Ranking Table“ (CIRT) erstellt, die die relevanten Unsicherheiten bei den unterschiedlichen Auslegungsstörfällen für Druckwasserreaktoren kategorisieren. Damit können Unsicherheitsanalysen basierend auf dem statistischen Verfahren nach S. S. Wilks durchgeführt werden. Schlussendlich werden die sicherheitsrelevanten Rechenergebnisse realistisch und vor allem mit einer hohen Zuverlässigkeit, im Vergleich zu einer herkömmlichen konservativen Berechnungsmethode, ermittelt.:1 Einleitung 2 Sicherheitsanforderungen an Kernkraftwerke 2.1 Einschluss radioaktiver Stoffe und Abschirmung ionisierender Strahlung 2.2 Störfallkategorien 2.3 Thermohydraulische Nachweiskriterien 3 Analysetechniken für die Durchführung von Störfalluntersuchungen 3.1 Das Störfallanalyseprogramm TRACE 3.2 Identifikation von Unsicherheiten bei Störfalluntersuchungen 3.3 Konservative Methode und Best-Estimate-Methode mit Unsicherheitsanalyse 3.4 Stand von Wissenschaft und Technik der wesentlichen Best-Estimate-Methoden 3.4.1 CSAU-Methode 3.4.2 UMAE-Methode 3.4.3 CIAU-Methode 3.4.4 GRS-Methode 3.4.5 ASTRU-Methode 4 Entwicklung der DYBESA-Methode für DWR-Störfalluntersuchungen 4.1 Thermohydraulische Phänomene beim Störfallablauf 4.2 Regressionsverfahren für die experimentelle Datenanalyse 4.2.1 Vertrauens- und Vorhersagebereich 4.2.2 Statistische Toleranzgrenzen 4.3 Identifikation von empirischen Beziehungen und deren Bewertung 4.4 Erzeugen und Kombinieren von geeigneten Stichproben 4.5 Programm zur Vorbereitung und Auswertung von Unsicherheitsanalysen 4.6 Modifikation des Störfallanalyseprogramms TRACE für die Berücksichtigung der programmspezifischen Unsicherheit 4.7 Verifikation der DYBESA-Methode 5 Ergebnisse 5.1 FEBA 5.1.1 Verifikation Rechenmodell 5.1.2 Verifikation CIRT 5.2 Marviken-Test-Station 5.2.1 Verifikation Rechenmodell 5.2.2 Verifikation CIRT 5.3 Druckwasserreaktor 5.3.1 Mittlerer Bruch einer Hauptkühlmittelleitung 5.3.2 Notstromfall 6 Diskussion und Ausblick 7 Zusammenfassung 8 Quellenverzeichnis 9 Anlagenverzeichnis

Nous avons étudiés des fils quantiques dopés de semi-conducteurs gravés par spectroscopie de diffusion Raman. Nous avons observés les excitations du gaz d'électrons. Celles-ci présentent des règles de sélection différentes de celles établies pour les systèmes bi-dimensionnels. Nous avons montré théoriquement qu'elles proviennent de la modification de la structure du champ électromagnétique local provoquée par la géométrie particulière des fils gravés. Pour cela nous avons dû calculer le champ local et l'introduire dans la section efficace de diffusion Raman pour en déduire les règles de sélection de toutes les excitations. Cela a permis de déterminer sans équivoque la nature des excitations qui sont des plasmons. Aucune excitations à une particule ni fluctuations de densité de spin n'a été observées. Nous avons étudié l'évolution continue des dispersions de ces plasmons lorsque la largeur du fil est réduite de 1 micromètre à 30 nm. Jusqu'à 60 nm, celles-ci sont en très bon accord avec les résultats d'un modèle hydrodynamique. Au dessous de 60 nm, la comparaison avec un modèle RPA s'impose. Le plasmon intra-bande dispersif est observé jusqu'à 45 nm, largeur en dessous de laquelle les spectres Raman sont dominés par des excitations localisées qui nécessitent une analyse ultérieure pour en déterminer clairement leur nature. Nous montrons à l'aide du modèle RPA que nous avons atteint la limite quantique pour un fil de largeur 55 nm. Une gamme étroite de fils dont les largeurs sont comprises entre 55 nm à 45 nm permet donc l'étude de gaz strictement unidimensionnel.<br />Nous avons cherché à déterminer la contribution de la forte illumination dans les conclusions précédentes. Nous avons utilisé pour cela la spectroscopie de magnéto-transmission infra-rouge qui ne modifie pas les conditions d'équilibre du gaz d'électrons. Une largeur critique de 130 nm a été extraite, en dessous de laquelle nous n'avons plus aucun signe de la présence d'électrons libres. La comparaison des mesures Raman et infra-rouge a permis l'établissement et la validation d'un modèle microscopique du potentiel de confinement présent dans les fils. Enfin nous avons fabriqués des échantillons de géométries plus complexes. L'observation et l'analyse par diffusion Raman des plasmons dans ces fils a montré que nous pouvions contrôler la géométrie du potentiel confinant les électrons et a mis en évidence des effets nouveaux tels que le repliement et le confinement de plasmons unidimensionnels.