Submitted by JÚLIO HEBER SILVA (julioheber@yahoo.com.br) on 2017-11-03T13:22:18Z No. of bitstreams: 2 Dissertação - Layara de Paula Sousa Santos - 2017.pdf: 2403100 bytes, checksum: 73792e08ef369454156c844fd7e57d55 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2017-11-06T09:47:01Z (GMT) No. of bitstreams: 2 Dissertação - Layara de Paula Sousa Santos - 2017.pdf: 2403100 bytes, checksum: 73792e08ef369454156c844fd7e57d55 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Made available in DSpace on 2017-11-06T09:47:01Z (GMT). No. of bitstreams: 2 Dissertação - Layara de Paula Sousa Santos - 2017.pdf: 2403100 bytes, checksum: 73792e08ef369454156c844fd7e57d55 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2017-09-29 / Faced with urban population growth and the importance of water as a limited natural resource, there is a need to implement techniques to reduce the operational costs of water distribution systems and ensure adequate supply. The optimization of pump operation can be used to meet the demands of consumption with a lower energy cost, in addition to maximizing hydraulic reliability. In this work, a hybrid optimization / simulation model was developed based on the multiobjective genetic algorithms and the EPANET hydraulic simulator. The NSGA II (Nondominated Sorting Genetic Algorithm II) method was used to optimize the operation of variable rotation pumps, that is, the decision variables of the problem were the rotation of the pumps for each hour throughout the day. A modification of the original EPANET hydraulic simulator, which does not correctly compute the efficiency of variable-speed pumps, was employed so that the power of each pump, and consequently the cost of electric power, was calculated correctly. The representation of the system in the model was done by means of the configuration of the hypothetical network called ANYTOWN in the EPANET and implementation of the Multiobjective Evolutionary Algorithm, determination of the penalty coefficients and determination of genetic parameters and operators (population, generation number, mutation probability and probability of recombination). The validity test of the developed model was obtained through simulations performed with the input data, including the patterns of variation of the speed of rotation of the pumps. Non-dominated solutions (Pareto Front) were obtained considering first the negative pressure penalty at the nodes and, subsequently, the negative pressure penalties at the nodes and the closure / shutdown of tubes and / or pumps. All points found represent optimal operating solutions for the system considering the period of the last 24 hours for calculating the objective functions. The results obtained for the two previously defined objectives demonstrate the effectiveness of the model, since mainly with adoption of penalty 2, presents adequate pressures at the nodes and adequate water level in the reservoir, with the consequent saving of electric energy and increased hydraulic reliability. / Diante do crescimento populacional urbano e importância da água como recurso natural limitado, verifica-se a necessidade de implementar técnicas com a finalidade de reduzir os custos operacionais dos sistemas de distribuição de água e garantir abastecimento adequado. A otimização da operação de bombas pode ser utilizada com o propósito de atender as demandas de consumo com um menor custo energético, além de maximizar a confiabilidade hidráulica. Neste trabalho, um modelo híbrido de otimização/simulação foi desenvolvido tendo como suporte os algoritmos genéticos multiobjetivo e o simulador hidráulico EPANET. O método NSGA II (Non-dominated Sorting Genetic Algorithm II) foi utilizado para a otimização da operação de bombas de rotação variável, ou seja, as variáveis de decisão do problema foram a rotação das bombas para cada hora ao longo do dia. Uma modificação do simulador hidráulico EPANET original, que não computa corretamente o rendimento de bombas de rotação variável, foi empregada para que as potências de cada bomba e, consequentemente o custo da energia elétrica, fossem calculadas corretamente. Foi realizada a representação do sistema no modelo, por meio da configuração da rede hipotética denominada ANYTOWN no EPANET e implementação do Algoritmo Evolucionário Multiobjetivo, determinação dos coeficientes de penalidade e determinação dos parâmetros e operadores genéticos (população, número de geração, probabilidade de mutação e probabilidade de recombinação). O teste de validade do modelo desenvolvido foi obtido por meio das simulações realizadas com os dados de entrada, incluindo os padrões de variação da velocidade de rotação das bombas. Conjuntos de soluções não dominadas (Frente Pareto) foram obtidos considerando-se primeiramente a penalidade de pressão negativa nos nós e, posteriormente, as penalidades de pressão negativa nos nós e fechamento/desligamento de tubos e/ou bombas. Todos os pontos encontrados representam soluções operacionais ótimas para o sistema considerando-se o período das últimas 24 horas para o cálculo das funções objetivo. Os resultados obtidos para os dois objetivos previamente definidos demonstram a eficácia do modelo, visto que principalmente com adoção da penalidade 2, apresenta pressões adequadas nos nós e nível adequado de água no reservatório, com a consequente economia de energia elétrica e aumento da confiabilidade hidráulica.

Submitted by Luciana Ferreira (lucgeral@gmail.com) on 2018-01-10T09:42:22Z No. of bitstreams: 2 Dissertação - Hélios Kárum de Oliveira Bastos - 2017.pdf: 2219804 bytes, checksum: ba853c18f7e7e2c65eb0a342d4a34640 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2018-01-10T09:42:46Z (GMT) No. of bitstreams: 2 Dissertação - Hélios Kárum de Oliveira Bastos - 2017.pdf: 2219804 bytes, checksum: ba853c18f7e7e2c65eb0a342d4a34640 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Made available in DSpace on 2018-01-10T09:42:46Z (GMT). No. of bitstreams: 2 Dissertação - Hélios Kárum de Oliveira Bastos - 2017.pdf: 2219804 bytes, checksum: ba853c18f7e7e2c65eb0a342d4a34640 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2017-08-24 / Multiobjective Optimization for feature and samples selection in double chromosome of integer representation and variable size for multivariate calibration} In several problems of regression, classification, prediction, approximation Optimization, the original data contain a large number of variables to obtain a better representation of the problem under consideration. However, a significant part of the variables may be irrelevant and redundant from the point of view of machine learning. Indeed, one of the challenges to be overcome is a selection of a subset of variables that has the best perform. One of the breakthroughs in this type of problem is the use of a multiobjective formulation that avoids the overlap of the model to the training data set. Another important point is the process of choosing the objects to be used in the learning stage. Generally, a selection of variables and treatment objects are treated separately and without dependence. This project proposes a multiobjective modeling to select variables and objects simultaneously using a genetic integer representation algorithm with variable size chromosomes. It is expected that a simultaneous selection of objects and variables on a multiobjective context produce better results in a traditional approach. As a case study this work utilized an analysis of near infrared (NIR) material on oil samples for the purpose of estimating the concentration of an interest properties such set was used in the competition conducted at the International Diffuse Reflectance Conference (IDRC) in the year 2014. / Em diversos problemas de regressão, classificação, previsão, aproximação e otimização, os dados originais contêm um grande número de variáveis introduzidas para se obter uma melhor representação do problema considerado. Entretanto, uma parte significativa destas variáveis podem ser irrelevantes e/ou redundantes do ponto de vista do aprendizado de máquina acerca do problema. Com efeito, um dos desafios a ser superados é a seleção de um subconjunto de variáveis que apresentem um melhor desempenho. Um dos avanços recentes neste tipo de problema está no uso de uma formulação multiobjetivo que evita o superajuste do modelo ao conjunto de dados de treinamento. Outro ponto importante refere-se ao processo de escolha adequada dos objetos a serem utilizados na etapa de aprendizado. Geralmente, a seleção de variáveis e de objetos de treinamento são tratados de forma separada e sem dependência. Este projeto propõe uma modelagem multiobjetivo para seleção de variáveis e objetos de forma simultânea utilizando-se de algoritmo genético de representação inteira com cromossomos de tamanho variáveis. Espera-se que a seleção simultânea de objetos e variáveis no contexto multiobjetivo produza melhores resultados em relação a abordagem tradicional. Como estudo de caso este trabalho utiliza dados obtidos por uma análise de material com ondas de infravermelho próximo (NIR) sobre amostras de petróleo com o propósito de estimar a concentração de uma propriedade de interesse, tal conjunto foi utilizado na competição realizada no International Diffuse Reflectance Conference (IDRC) (\url{http://cnirs.clubexpress.com/content.aspx?page_id=22&club_id=409746&module_id=19 0211}), no ano de 2015.

A crescente demanda por biocombustíveis e a concorrência dos combustíveis fósseis torna necessária a otimização dos meios já existentes e o desenvolvimento de novas tecnologias para produção de biocombustíveis, principalmente em projetos envolvendo biorreatores e fotobiorreatores (FBR) industriais. A dinâmica dos fluidos computacional (CFD) vem sendo utilizada em vários trabalhos para o estudo de parâmetros fluidodinâmicos que podem influenciar no rendimento dos processos químicos envolvidos, como tensão de cisalhamento, perfis de velocidade, tempo de residência e a influência da geometria sobre esses parâmetros. Contudo, não existe ainda um número abrangente de trabalhos que utilize técnicas de otimização acopladas com a resolução numérica do problema fluidodinâmico. Em alguns estudos, algoritmos de otimização são utilizados para determinar os melhores coeficientes das reações químicas. No entanto, não há estudos, até o momento, que reportem a otimização multiobjetivo simultânea dos parâmetros geométricos e do escoamento aplicados a equipamentos da indústria sucro-energética. Neste contexto, o presente trabalho de pesquisa tem como objetivo contribuir para o avanço científico e tecnológico através da implementação de um software aberto (PyCFD-O) que permita o acoplamento CFD-otimização e o desenvolvimento das bases de uma metodologia de projeto otimizado bem como de operação de biorreatores e FBRs de escala industrial. O PyCFD-O foi testado em dois estudos de caso que podem ser estendidos a um fermentador contínuo e um FBR. Os parâmetros geométricos de ambos os reatores foram otimizados de forma a minimizar simultaneamente a tensão de cisalhamento e a variância da distribuição do tempo de residência. O software PyCFD-O mostrou-se robusto, revelando que o processo global de otimização realiza de fato a busca pela fronteira de Pareto. Além da obtenção das geometrias otimizadas, também foram discutidos a influência dos parâmetros geométricos na hidrodinâmica do escoamento em ambos os casos. / The growing demand for biofuels and its competition with fossil fuels create the need to optimize the existing resources and development of new technologies for production of biofuels, particularly in projects involving industrial bioreactors and photobioreactors (PBR). Computational fluid dynamics (CFD) has been used in several studies for the study of fluid dynamics parameters that can influence the performance of the chemical process involved, such as shear stress, velocity profiles, residence time and the influence of these parameters on the reactor geometry. However, there are lacks of studies that utilize optimization techniques coupled with the numerical resolution of the fluid dynamic problem. The use of optimization algorithms has been reported in some cases, but there have not been reports on studies combining the optimization of flow parameters and multiobjective algorithms to choose ideal geometric parameters applied to the equipment of the sugar-energy industry. In this context, this research project aims to contribute to the advancement of scientific and technological knowledge trhough the implementation of open source software (PyCFD-O) for the CFD-optimization coupling and the development of the bases of a methodology for optimal design and operation of industrial scale bioreactors and PBR. The PyCFD-O software was tested in two case studies with characteristics that can be extended to a continuos fermenter and PBR. The geometric parameters of both reactors were simultaneously optimized in order to minimize the shear stress and the variance of residence time distribuition. The PyCFD-O software showed robustness, revealing that overall optimization process actually performs the search of Pareto frontier. In addition to the geometry optimization, the influence of the geometrical parameters of the hydrodynamic of the flow was discussed in both case studies.

Submitted by Geandra Rodrigues (geandrar@gmail.com) on 2018-07-19T15:32:30Z No. of bitstreams: 1 felipegomesduque.pdf: 3507645 bytes, checksum: d7e6e84ea872f86ac2bd64caa361fce0 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2018-07-23T15:26:16Z (GMT) No. of bitstreams: 1 felipegomesduque.pdf: 3507645 bytes, checksum: d7e6e84ea872f86ac2bd64caa361fce0 (MD5) / Made available in DSpace on 2018-07-23T15:26:16Z (GMT). No. of bitstreams: 1 felipegomesduque.pdf: 3507645 bytes, checksum: d7e6e84ea872f86ac2bd64caa361fce0 (MD5) Previous issue date: 2018-04-12 / O presente trabalho propõe uma metodologia de planejamento de medição em Sistemas de Distribuição de Energia Elétrica (SDE) e um novo método para estimação de estados destes sistemas. Para tanto, a técnica metaheurística de otimização bio-inspirada denominada Modified Monkey Search (MMS) é proposta para alocação ótima de medidores inteligentes e unidades de medição fasorial. O modelo de otimização é multiobjetivo e visa a maximização da eficácia do processo de estimação de estados com o custo mínimo de investimento em sistemas de medição. O método de Pareto é associado ao algoritmo MMS para o tratamento adequado destes objetivos conflitantes considerando-se custos reais associados aos equipamentos de medição. Adicionalmente, um novo método de estimação de estados baseado na modelagem de um Fluxo de Potência Ótimo (FPO) modificado é proposto, cuja resolução é dada pelo Método de Pontos Interiores (MPI). O algoritmo MMS determina as variáveis discretas associadas aos tipos de equipamentos de medição, bem como aos locais de instalação dos mesmos no SDE. Estudos são realizados para comparar a nova metodologia de estimação de estados proposta com uma metodologia tradicional, bem como para comparar os resultados da metaheurística de otimização aplicada ao problema com outras técnicas desenvolvidas para esta finalidade. Os estudos são conduzidos com sistemas da literatura, além de um sistema real de médio porte de uma concessionária brasileira. / The present work proposes an approach for planning the measurement locations in Electric Distribution Systems (EDS) and a new method for static state estimation. The bio-inspired meta-heuristic optimization technique called Modified Monkey Search (MMS) is proposed for optimal allocation of smart meters and phasor measurement units. The optimization model is multiobjective and aims at maximizing the efficiency of the state estimation process with minimum measurement investment costs. The Pareto’s method is associated with the MMS algorithm for handling the conflicting objectives in a suitable manner by considering real costs related to measurement equipments. In addition, a new method for static state estimation based on the modeling of a modified Optimal Power Flow (OPF) is proposed, whose solution is given by the Interior Point Method (IPM). The MMS algorithm determines the discrete variables related to types and location of measurement equipments in the system. Studies are made to compare the new approach for static state estimation with a traditional method, as well as to compare the results from the meta-heuristic optimization applied to the problem with existing techniques. The studies are performed using systems from the literature, as will as a practical medium size distribution network from a Brazilian utility.

Flooding is an important concern for the UK, as evidenced by the many extreme flooding events in the last decade. Improved flood risk intervention strategies are therefore highly desirable. The application of hydroinformatics tools, and optimisation algorithms in particular, which could provide guidance towards improved intervention strategies, is hindered by the necessity of performing flood modelling in the process of evaluating solutions. Flood modelling is a computationally demanding task; reducing its impact upon the optimisation process would therefore be a significant achievement and of considerable benefit to this research area. In this thesis sophisticated multi-objective optimisation algorithms have been utilised in combination with cutting-edge flood-risk assessment models to identify least-cost and most-benefit flood risk interventions that can be made on a drainage network. Software analysis and optimisation has improved the flood risk model performance. Additionally, artificial neural networks used as feature detectors have been employed as part of a novel development of an optimisation algorithm. This has alleviated the computational time-demands caused by using extremely complex models. The results from testing indicate that the developed algorithm with feature detectors outperforms (given limited computational resources available) a base multi-objective genetic algorithm. It does so in terms of both dominated hypervolume and a modified convergence metric, at each iteration. This indicates both that a shorter run of the algorithm produces a more optimal result than a similar length run of a chosen base algorithm, and also that a full run to complete convergence takes fewer iterations (and therefore less time) with the new algorithm.

Submitted by Renata Lopes (renatasil82@gmail.com) on 2016-01-15T14:16:25Z No. of bitstreams: 1 denisemanueldacostavargas.pdf: 16589539 bytes, checksum: 44a0869db27ffd5f8254f85fb69ab78c (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2016-01-25T17:40:31Z (GMT) No. of bitstreams: 1 denisemanueldacostavargas.pdf: 16589539 bytes, checksum: 44a0869db27ffd5f8254f85fb69ab78c (MD5) / Made available in DSpace on 2016-01-25T17:40:31Z (GMT). No. of bitstreams: 1 denisemanueldacostavargas.pdf: 16589539 bytes, checksum: 44a0869db27ffd5f8254f85fb69ab78c (MD5) Previous issue date: 2015-11-05 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Problemas de Otimização Multiobjetivo (POMs) com restrições são frequentes em diversas áreas das ciências e engenharia, entre elas a Otimização Estrutural (OE). Apesar da Evolução Diferencial (ED) ser uma metaheurística muito atraente na resolução de problemas do mundo real, há uma carência na literatura de discussões sobre o desempenho em POMs de OE. Na sua grande maioria os problemas de OE apresentam restrições. Nesta tese utiliza-se uma técnica para o tratamento de restrições chamada de APM (Adaptive Penalty Method) que tem histórico de bons resultados quando aplicada em problemas monobjetivo de OE. Pelo potencial da ED na resolução de problemas do mundo real e da técnica APM em OE, juntamente com a escassez de trabalhos envolvendo esses elementos em POMs de OE, essa tese apresenta um estudo de um algoritmo bem conhecido de ED acoplado à técnica APM nesses problemas. Experimentos computacionais considerando cenários sem e com inserção de informações de preferência do usuário foram realizados em problemas com variáveis continuas e discretas. Os resultados foram comparados aos encontrados na literatura, além dos obtidos pelo algoritmo que representa o estado da arte. Comparou-se também os resultados obtidos pelo mesmo algoritmo de ED adotado, porém sem ser acoplado à técnica APM, objetivando investigar sua influência no desempenho da combinação proposta. As vantagens e desvantagens do algoritmo proposto em cada cenário são apresentadas nessa tese, além de sugestões para trabalhos futuros. / Multiobjective Optimization Problems (MOPs) with constraints are common in many areas of science and engineering, such as Structural Optimization (SO). In spite of Differential Evolution (DE) being a very attractive metaheuristic in real-world problems, no work was found assessing its performance in SO MOPs. Most OE problems have constraints. This thesis uses the constraint handling technique called Adaptive Penalty Method (APM) that has a history of good results when applied in monobjective problems of SO. Due to the potential of DE in solving real world problems and APM in SO problems, and also with the lack of studies involving these elements in SO MOPs, this work presents a study of a well-known DE algorithm coupled to the APM technique in these problems. Computational experiments considering scenarios with and without inclusion of user preference information were performed in problems with continuous and discrete variables. The results were compared with those in the literature, in addition to those obtained by the algorithm that represents the state of the art. They were also compared to the results obtained by the same DE algorithm adopted, but without the APM technique, aiming at investigating the influence of the APM technique in their performance. The advantages and disadvantages of the proposed algorithm in each scenario are presented in this research, as well as suggestions for future works.

Constructing a road is a complex process that may be represented as a series of correlated steps, from the planning to the construction and usage of the new road. At the heart of this process, the preliminary and detailed design stages are key elements that will ensure the quality and the adequacy of the final solution regarding the constraints and objectives of the project. In particular, infrastructure layout and design will have a strong impact on the global performances of the road in operational conditions. Among them, road safety, mobility, environment preservation, noise pollution limitation, economic feasibility and viability of the project, or even its socio-economic impact at the local level. Consequently, it is crucial to offer engineers and road planners some tools and methods that may assist them in designing and selecting the most efficient solutions considering the distinctive features of each design problem. In this work, a multicriteria analysis methodology is developed to carry out an integrated and preventive assessment of road projects at the design stage by considering both their safety performances and some economic and environmental aspects. Its purpose is to support design engineers in the analysis of their projects and the identification of innovative, consistent and effective solutions. The proposed methodology is composed of two main research frameworks. On the one hand, the road design problem is addressed by focusing successively on the structuring of the multicriteria problem, the identification of the approximate set of non-dominated solutions using a genetic algorithm (based on NSGA-II), and the application of the methodology to a real road design project. On the other hand, the methodological development of a multicriteria interval clustering model was performed (based on PROMETHEE). Due to the applicability of this model to the studied problem, the interactions between the two frameworks are also analysed. / Doctorat en Sciences de l'ingénieur et technologie / The present PhD thesis is an aggregation of published contributions related to the application of multicriteria analysis to the evaluation of road projects at the design stage. The aim of the two introductory chapters is to offer a synthesised and critical presentation of the scientific contributions that constitute the PhD thesis. The complete version of the journal articles and preprints are found in Chapters 3 to 6. In the appendices, we also provide reprints of conference papers that are usually related to one of the main contributions of the thesis. / info:eu-repo/semantics/nonPublished

This thesis deals with radio network design for a chosen part of a map. Here map is represented by digital map file, which was created within the project DEM. First step is to calculate distances between points in chosen map. With help of optimization algorithms, appropriate position of transceiver in the map and parameters of radio systems are determined, also final coverage by radio signal, represented by intensity of electric field or received power in whole map. The optimization algorithm is used to find the best solution in terms of input parameters (e.g. power of transmitter, height of mast) and resulting coverage of land by radio signal.

[pt] Problemas reais de engenharia mecânica podem compreender tarefas de i) otimização multi-objetivo (MO) ou ii) regressão, classificação e predição. Os métodos baseados em inteligência artificial (AI) são bastante difundidos na resolução desses problemas por i) demandarem menor custo computacional e informações do domínio do problema para a resolução de uma MO, quando comparados com métodos de programação matemática, por exemplo; e ii) apresentarem melhores resultados com estrutura mais simples, adaptabilidade e interpretabilidade, em contraste com outros métodos. Sendo assim, o presente trabalho busca i) otimizar um controle proporcional-integral-derivativo (PID) aplicado a um sistema de frenagem anti-travamento de rodas (ABS) e o projeto de trocadores de calor de placas aletadas (PFHE) e casco-tubo (STHE) através de métodos de otimização baseados AI, buscando o desenvolvimento de novas versões dos métodos aplicados, e.g. multi-objective salp swarm algorithm (MSSA) e multi-objective heuristic Kalman algorithm (MOHKA), que melhorem a performance da otimização; ii) desenvolver um sistema de detecção de vazamento em dutos (LDS) sensível ao roubo de combustível a partir do treinamento de árvores de decisão (DTs) com features baseadas no tempo e na análise de componentes principais (PCA), ambas exraídas de dados de transiente de pressão de operação normal do duto e de roubo de combustível; iii) constituir um guia de aplicação para problemas de MO de controle e projeto, processo de extração de features e treinamento de classificadores baseados em aprendizado de máquina (MLCs), através de aprendizado supervisionado; e, por fim iv) demonstrar o potencial das técnicas baseadas em AI. / [en] Real-world mechanical engineering problems may comprise tasks of i) multi-objective optimization (MO) or ii) regression, classification and prediction. The use of artificial intelligence (AI) based methods for solving these problems are widespread for i) demanding less computational cost and problem domain information to solve the MO, when compared with mathematical programming for an example; and ii) presenting better results with simpler structure, adaptability and interpretability, in contrast to other methods. Therefore, the present work seeks to i) optimize a proportional-integral-derivative control (PID) applied to an anti-lock braking system (ABS) and the heat exchanger design of plate-fin (PFHE) and shell-tube (STHE) types through AI based optimization methods, seeking to develop new versions of the applied methods, e.g. multi-objective salp swarm algorithm (MSSA) and multi-objective heuristic Kalman algorithm (MOHKA), which enhance the optimization performance; ii) develop a pipeline leak detection system (LDS) sensitive to fuel theft by training decision trees (DTs) with features based on time and principal component analysis (PCA), both extracted from pressure transient data of regular pipeline operation and fuel theft; iii) constitute an application guide for control and design MO problems, feature extraction process and machine learning classifiers (MLCs) training through supervised learning; and, finally, iv) demonstrate the potential of AI-based techniques.

工学設計とは,複数の性能要求を満たすべく,膨大な数の設計変数の最適値を求め,自らが望む設計を選択していく,設計探査と意思決定のプロセスである. 本研究では,これら２つの主要課題に対して,多様かつ優れた解を高速に得るための「設計空間探査」と高次元かつ膨大な設計候補群から主要な設計パターンを抽出し,その特徴を定量的に分析するための「設計モード解析」の技術を開発し,その有効性を示した. / In engineering design, we try to find better solutions that satisfy many design requirements. Once the designs have been found, we choose preferable one. Engineering design is the mixed procedure of design exploration and decision making. This study proposes the effective solution for each engineering process: (1) a novel optimization algorithm to rapidly derive better designs, and (2) "design mode analysis" which enables us to extract representative design patterns from the huge number of design examples. Effectiveness of the proposed method were verified in real-world problems. / 博士(工学) / Doctor of Philosophy in Engineering / 同志社大学 / Doshisha University