本論文針對多廠區訂單分配（Multi-plant order allocation）問題進行探討，此問題模式下企業擁有多間製造不同產品之工廠，且生產成本、產能、運送成本等也各自不同，因此這些因素都必須納入訂單分配時的考量。研究中同時考量三個目標：製造成本、配送前置時間和工廠平均產能利用率之均衡性，利用層級分析法（AHP）將三者進行結合，以達到多目標規劃。除了提出此模型架構外，並以基因演算法（Genetic Algorithm）結合層級分析法進行問題的求解，以達到最佳的分配方式，而為了加強求解的品質與效率，利用禁忌搜尋法（Tabu Search）來改善演化過程中，對於產生不可行解的處理方式。在研究最後，將計算結果與過去研究成果作比較，顯示採用基因演算法混合禁忌搜尋法，在求解多廠區訂單分配問題時，可以得到較佳的結果。

By definition, all projects are unique / however R&amp / D projects have specific characteristics that make them harder to manage. The project management methodology applied to R&amp / D projects may show differences due to the categorization of them. But if there exists a categorization of projects, one can analyze the properties of the project classes and then manage similar projects similarly. In this study, the R&amp / D projects of a main military electronics company of Turkey, are analyzed. Tun&ccedil / (2004) has developed a methodology for clustering the projects of this electronics company. Continuing from his studies, a methodology for determining the class of a new project of this electronics company is developed. For defining the projects in a project space, a Project Identification Card (PIC) is developed. The measurement scale of the PIC is constructed by using the absolutemeasurement Analytic Hierarchy Process. A clustering Tabu Search algorithm is generated for using in the sensitivity analyses of the clusters to projects. And a methodology for determining the cluster of a new project is developed.

Esta pesquisa consiste na solução do problema inverso de transferência radiativa para um meio participante (emissor, absorvedor e/ou espalhador) homogêneo unidimensional em uma camada, usando-se a combinação de rede neural artificial (RNA) com técnicas de otimização. A saída da RNA, devidamente treinada, apresenta os valores das propriedades radiativas [ω, τ0, ρ1 e ρ2] que são otimizadas através das seguintes técnicas: Particle Collision Algorithm (PCA), Algoritmos Genéticos (AG), Greedy Randomized Adaptive Search Procedure (GRASP) e Busca Tabu (BT). Os dados usados no treinamento da RNA são sintéticos, gerados através do problema direto sem a introdução de ruído. Os resultados obtidos unicamente pela RNA, apresentam um erro médio percentual menor que 1,64%, seria satisfatório, todavia para o tratamento usando-se as quatro técnicas de otimização citadas anteriormente, os resultados tornaram-se ainda melhores com erros percentuais menores que 0,04%, especialmente quando a otimização é feita por AG. / This research consists in the solution of the inverse problem of radiative transfer for a participating media (emmiting, absorbing and/or scattering) homogeneous one-dimensional in one layer, using the combination of artificial neural network (ANN), with optimization techniques. The output of the ANN, properly trained presents the values of the radiative properties [w, to, p1 e p2] that are optimized through the following techniques: Particle Collision Algorithm (PCA), Genetic Algorithm (GA), Greedy Randomized Adaptive Search Procedure (GRASP) and Tabu Search (TS). The data used in the training are synthetics, generated through the direct problem without the introduction of noise. The results obtained by the (ANN) alone, presents an average percentage error minor than 1,64%, what it would be satisfying, however, for the treatment using the four techniques of optimization aforementioned, the results have become even better with percentage errors minor than 0,03%, especially when the optimization is made by the GA.

Made available in DSpace on 2014-12-17T15:48:07Z (GMT). No. of bitstreams: 1 CamilaNOT_DISSERT.pdf: 4321465 bytes, checksum: d103bd2da647997e8dfd0a8784c2060d (MD5) Previous issue date: 2013-02-01 / Nonogram is a logical puzzle whose associated decision problem is NP-complete. It has applications in pattern recognition problems and data compression, among others. The puzzle consists in determining an assignment of colors to pixels distributed in a N  M matrix that satisfies line and column constraints. A Nonogram is encoded by a vector whose elements specify the number of pixels in each row and column of a figure without specifying their coordinates. This work presents exact and heuristic approaches to solve Nonograms. The depth first search was one of the chosen exact approaches because it is a typical example of brute search algorithm that is easy to implement. Another implemented exact approach was based on the Las Vegas algorithm, so that we intend to investigate whether the randomness introduce by the Las Vegas-based algorithm would be an advantage over the depth first search. The Nonogram is also transformed into a Constraint Satisfaction Problem. Three heuristics approaches are proposed: a Tabu Search and two memetic algorithms. A new function to calculate the objective function is proposed. The approaches are applied on 234 instances, the size of the instances ranging from 5 x 5 to 100 x 100 size, and including logical and random Nonograms / O Nonograma ? um jogo l?gico cujo problema de decis?o associado ? NP-completo. Ele possui aplica??o em problemas de identifica??o de padr?es e de compacta??o de dados, dentre outros. O jogo consiste em determinar uma aloca??o de cores em pixels distribu?dos em uma matriz N  M atendendo restri??es em linhas e colunas. Um Nonograma ? codificado atrav?s de vetores cujos elementos especificam o n?mero de pixels existentes em cada coluna e linha de uma figura, sem especificar suas coordenadas. Este trabalho apresenta abordagens exatas e heur?sticas para solucionar o Nonograma. A Busca em Profundidade foi uma das abordagens exatas escolhida, por ser um exemplo t?pico de algoritmo de for?a bruta de f?cil implementa??o. Outra abordagem exata implementada foi baseada no algoritmo Las Vegas, atrav?s do qual se pretende investigar se a aleatoriedade introduzida pelo algoritmo Las Vegas traria algum benef?cio em rela??o ? Busca em Profundidade. O Nonograma tamb?m ? transformado em um Problema de Satisfa??o de Restri??es. Tr?s abordagens heur?sticas s?o propostas: uma Busca Tabu e dois algoritmos Mem?tico. Uma nova abordagem para o c?lculo da fun??o objetivo ? proposta neste trabalho. As abordagens s?o testadas em 234 casos de teste de tamanho entre 5 x 5 e 100 x 100, incluindo Nonogramas l?gicos e aleat?rios

No description available.

Les travaux présentes dans cette thèse s'inscrivent dans le cadre des méthodes d'évacuation des populations. Ils visent à étudier les capacités et modéliser le problème d'évacuation (blessés, sinistrés, enfants, personnes agées, etc.) dans une situation de crise (attentats terroristes, catastrophes naturelles, etc.) et développer des méthodes d'aide à la décision tout en proposant une meilleure planification et des plans optimaux d'évacuation des populations de la zone de crise vers les centres hospitaliers.Notre travail consiste à résoudre le problème d'évacuation de blessés dans des zones de crise avec une nouvelle vision qui consiste à optimiser le temps de transport et par conséquent sauver le maximum des personnes touchées par cette crise d'une façon dynamique, efficace et rapide pour minimiser la perte humaine. / The work presented in this thesis is part of human evacuation methods. It aims to study the capacities, model the evacuation problem (wounded, victims, children, elderly, etc.) in a crisis situation (terrorist attacks, natural disasters, etc.) and to develops methods for decision making while proposing better planning and optimal evacuation plans for populations from the crisis zone to hospitals.Our job is to solve the wounded evacuation problem in crisis zone with a new vision that optimizes the transport time and thus saving the maximum of causalities in a dynamic, efficient and fast way in order to minimize human loss.

Nous considérons dans cette thèse un problème d’ordonnancement de flow-shop de permutation où un ensemble de travaux doit être ordonnancé sur un ensemble de machines. Les travaux doivent être ordonnancés sur les machines dans le même ordre. L’objectif est de minimiser le retard total. Nous proposons des algorithmes heuristiques et des nouvelles matheuristiques pour ce problème. Les matheuristiques sont un nouveau type d’algorithmes approchés qui ont été proposés pour résoudre des problèmes d’optimisation combinatoire. Les méthodes importent de la résolution exacte au sein des approches (méta) heuristiques. Ce type de méthode de résolution a reçu un grand intérêt en raison de leurs très bonnes performances pour résoudre des problèmes difficiles. Nous présentons d’abord les concepts de base d’un problème d’ordonnancement. Nous donnons aussi une brève introduction à la théorie de l’ordonnancement et nous présentons un panel de méthodes de résolution. Enfin, nous considérons un problème où un flow shop de permutation à m-machine et un problème de tournées de véhicules sont intégrés, avec pour objectif la minimisation de la somme des retards. Nous proposons un codage direct d’une solution et une méthode de voisinage. Les résultats montrent que l’algorithme Tabou améliore grandement la solution initiale donnée par EDD et où chaque voyage ne délivre qu’un travail. / We consider in this thesis a permutation flow shop scheduling problem where a set of jobs have to be scheduled on a set of machines. The jobs have to be processed on the machines in the same order. The objective is to minimize the total tardiness. We propose heuristic algorithms and many new matheuristic algorithms for this problem. The matheuristic methods are a new type of approximated algorithms that have been proposed for solving combinatorial optimization problems. These methods embed exact resolution into (meta)heuristic approaches. This type of resolution method has received a great interest because of their very good performances for solving some difficult problems. We present the basic concepts and components of a scheduling problem and the aspects related to these components. We also give a brief introduction to the theory of scheduling and present an overview of resolution methods. Finally, we consider a problem where m-machine permutation flow shop scheduling problem and a vehicle routing problem are integrated and the objective is to minimize the total tardiness. We introduce a direct coding for a complete solution and a Tabu search for finding a sequence and trips. The results show that the TS greatly improves the initial solution given by EDD heuristic where each trip serves only one job at a time.

Orientador: José Roberto Sanches Mantovani / Resumo: Neste trabalho propõe-se o desenvolvimento de uma ferramenta computacional para o planejamento da operação de curto prazo de sistemas de distribuição com geração distribuída (GD) considerando uma abordagem probabilística. Uma modelagem sequencial formulada com base na perspectiva das companhias de distribuição (DisCos) é proposta. As decisões operacionais da DisCo são inicialmente otimizadas no estágio de operação day-ahead (DA) e, então, na operação real-time (RT). A operação DA visa maximizar a diferença entre a energia vendida aos consumidores e as compras realizadas no mercado de eletricidade atacadista e da GD, ou seja, os lucros. No estágio RT, busca-se a minimização dos ajustes necessários para acomodar os desvios das quantidades previstas no planejamento DA. Modelos de cargas dependentes de tensão e restrições relacionadas à demanda são explicitamente formulados. A rede é representada através de equações de fluxo de potência AC completo. Propõe-se ainda a incorporação de um mecanismo para precificação nodal de potência reativa. Os modelos resultantes são caracterizados como programas de otimização matemática multiperíodo de grande porte não lineares e não convexos com variáveis contínuas e discretas. Um algoritmo pseudodinâmico baseado na meta-heurística Busca Tabu (BT) é proposto para solução do problema resultante de maneira eficaz, sem linearizações. Os resultados obtidos para alimentadores de distribuição de 69 e 135 barras ilustram a eficiência da metodologia pro... (Resumo completo, clicar acesso eletrônico abaixo) / Doutor

La conception des filtres a réponse impulsionnelle finie (RIF) peut être formulée comme un problème d'optimisation non linéaire réputé pour être difficile sa résolution par les approches conventionnelles. Afin d'optimiser la conception des filtres RIF, nous explorons plusieurs méthodes stochastiques capables de traiter de grands espaces. Nous proposons un nouvel algorithme génétique dans lequel certains concepts innovants sont introduits pour améliorer la convergence et rendre son utilisation plus facile pour les praticiens. Le point clé de notre approche découle de la capacité de l'algorithme génétique (AG) pour adapter les opérateurs génétiques au cours de la vie génétique tout en restant simple et facile à mettre en oeuvre. Ensuite, l’optimisation par essaim de particules (PSO) est proposée pour la conception de filtres RIF. Finalement, un algorithme génétique hybride (HGA) est proposé pour la conception de filtres numériques. L'algorithme est composé d'un processus génétique pur et d’une approche locale dédiée. Notre contribution vise à relever le défi actuel de démocratisation de l'utilisation des AG’s pour les problèmes d’optimisation. Les expériences réalisées avec différents types de filtres mettent en évidence la contribution récurrente de l'hybridation dans l'amélioration des performances et montrent également les avantages de notre proposition par rapport à d'autres approches classiques de conception de filtres et d’autres AG’s de référence dans ce domaine d'application. / The design of finite impulse response (FIR) filters can be formulated as a non-linear optimization problem reputed to be difficult for conventional approaches. In order to optimize the design of FIR filters, we explore several stochastic methodologies capable of handling large spaces. We propose a new genetic algorithm in which some innovative concepts are introduced to improve the convergence and make its use easier for practitioners. The key point of our approach stems from the capacity of the genetic algorithm (GA) to adapt the genetic operators during the genetic life while remaining simple and easy to implement. Then, the Particle Swarm Optimization (PSO) is proposed for FIR filter design. Finally, a hybrid genetic algorithm (HGA) is proposed for the design of digital filters. The algorithm is composed of a pure genetic process and a dedicated local approach. Our contribution seeks to address the current challenge of democratizing the use of GAs for real optimization problems. Experiments performed with various types of filters highlight the recurrent contribution of hybridization in improving performance. The experiments also reveal the advantages of our proposal compared to more conventional filter design approaches and some reference GAs in this field of application.

Esta pesquisa consiste na solução do problema inverso de transferência radiativa para um meio participante (emissor, absorvedor e/ou espalhador) homogêneo unidimensional em uma camada, usando-se a combinação de rede neural artificial (RNA) com técnicas de otimização. A saída da RNA, devidamente treinada, apresenta os valores das propriedades radiativas [ω, τ0, ρ1 e ρ2] que são otimizadas através das seguintes técnicas: Particle Collision Algorithm (PCA), Algoritmos Genéticos (AG), Greedy Randomized Adaptive Search Procedure (GRASP) e Busca Tabu (BT). Os dados usados no treinamento da RNA são sintéticos, gerados através do problema direto sem a introdução de ruído. Os resultados obtidos unicamente pela RNA, apresentam um erro médio percentual menor que 1,64%, seria satisfatório, todavia para o tratamento usando-se as quatro técnicas de otimização citadas anteriormente, os resultados tornaram-se ainda melhores com erros percentuais menores que 0,04%, especialmente quando a otimização é feita por AG. / This research consists in the solution of the inverse problem of radiative transfer for a participating media (emmiting, absorbing and/or scattering) homogeneous one-dimensional in one layer, using the combination of artificial neural network (ANN), with optimization techniques. The output of the ANN, properly trained presents the values of the radiative properties [w, to, p1 e p2] that are optimized through the following techniques: Particle Collision Algorithm (PCA), Genetic Algorithm (GA), Greedy Randomized Adaptive Search Procedure (GRASP) and Tabu Search (TS). The data used in the training are synthetics, generated through the direct problem without the introduction of noise. The results obtained by the (ANN) alone, presents an average percentage error minor than 1,64%, what it would be satisfying, however, for the treatment using the four techniques of optimization aforementioned, the results have become even better with percentage errors minor than 0,03%, especially when the optimization is made by the GA.