Nos travaux concernent la mise en œuvre de méthodologies pour la résolution de problèmes d’ordonnancement en industries agroalimentaires. Trois nouvelles approches basées sur les algorithmes génétiques, sont proposées pour la résolution de problèmes d’ordonnancement multi-objectifs : les algorithmes génétiques séquentiels (SGA), les algorithmes génétiques parallèles (PGA) et les algorithmes génétiques parallèles séquentiels (PSGA). Deux approches coopératives multi-objectifs en mode relais, SH_GA/TS et SH_GA/SA, hybridant toutes les deux des métaheuristiques de haut niveau, sont par la suite proposées. Un algorithme évolutionnaire et un algorithme de recherche locale sont, dans ce cas exécutés séquentiellement / The purpose of our works is the implementation of methodologies for the resolution of the agro-food industry scheduling problem. Three new approaches based on genetic algorithms are proposed to solve multi-objectives scheduling problems: sequential genetic algorithms (SGA), parallel genetic algorithms (PGA) and parallel sequential genetic algorithms (PSGA). Two high-level hybrid algorithms, SH_GA/TS et SH_GA/SA, are also proposed. The purpose in this hybridization is to benefit the exploration of the solution space by a population of individuals with the exploitation of solutions through a smart search of the local search algorithm

The Vehicle Routing Problem has been around for more than 50 years and has been of major interest to the operations research community. The VRP pose a complex problem with major benefits for the industry. In every supply chain transportation occurs between customers and suppliers. In this thesis, we analyze the use of a multiple pheromone trial in using Ant Systems to solve the VRP. The goal is to find a reasonable solution for data environments of derivatives of the basic VRP. An adaptive object model approach is followed to allow for additional constraints and customizable cost functions. A parallel method is used to improve speed and traversing the solution space. The Ant System is applied to the local search operations as well as the data objects. The Tabu Search method is used in the local search part of the solution. The study succeeds in allowing for all of the key performance indicators, i.e. efficiency, effectiveness, alignment, agility and integration for an IT system, where the traditional research on a VRP algorithm only focuses on the first two. / Thesis (PhD)--University of Pretoria, 2010. / Industrial and Systems Engineering / unrestricted

Orientador: Rubén Augusto Romero Lazaro / Resumo: O problema de reconfiguração de sistemas de distribuição de energia elétrica consiste em encontrar uma configuração radial por meio da permutação do estado das chaves (abertura ou fechamento) dos ramos de um sistema elétrico. O objetivo é de se alcançar a minimização das perdas elétricas. Cada configuração radial só é considerada factível se respeitar certas restrições operacionais como o limite de tensão nas barras e os limites de correntes nos circuitos. O modelo tratado neste trabalho apresenta explosão combinatória e difícil tratabilidade por meio de métodos convencionais de otimização. O problema, computacionalmente falando, é considerado Não-Polinomial Completo (NPC), pois não possui uma resposta em tempo polinomial a partir de uma entrada definida. Neste trabalho são apresentadas três técnicas meta-heurísticas para se tratar o problema de reconfiguração de sistemas de distribuição de energia elétrica, totalmente diferentes entre uma e outra, atuando em conjunto, para somente um nível de demanda, no intuito de se encontrar a topologia ótima, com o objetivo de se minimizar as perdas elétricas ativas. Além disso, propôs-se modificar o paradigma clássico de implementação estático deste tipo de problema para o paradigma de programação dinâmica por meio de árvores com filhos variados a fim de que a estrutura de dados utilizada representasse fielmente um sistema de distribuição de energia elétrica na memória do computador. As meta-heurísticas implementadas foram a Greedy Rand... (Resumo completo, clicar acesso eletrônico abaixo) / Doutor

No description available.

Detta arbete syftar till att undersöka om det är möjligt att med en heuristisk metod skapa giltiga lösningar till ett problem vid planering av stränggjutningsproduktion på SSAB. Planeringsproblemet uppstår när stål av olika sorter ska gjutas under samma dag. Beroende på i vilken ordning olika kundordrar av stål gjuts uppstår spill av olika storlek. Detta spill ska minimeras och tidigare arbete har genomförts på detta problem och resulterat i en matematisk modell för att skapa lösningar till problemet. Det tar i praktiken lång tid att hitta bra lösningar med modellen och frågeställningen är om det går att göra detta med en heuristisk metod för att kunna generera bra lösningar snabbare. Med inspiration från Variable Neighbourhood Search, Simulated Annealing och tabusökning har heuristiker skapats, implementerats och utvärderats mot den matematiska modellen. En av heuristikerna presterar bättre än den matematiska modellen gör på 10 minuter. Matematiska modellens resultat efter 60 minuter körtid är bättre än den utvecklade heuristiken, men resultaten är nära varandra. Körtiden för heuristiken tar signifikant mindre tid än 10 minuter. / This study aims to investigate if it is possible to use a heuristic method to create feasible solution in a Cast Batching Problem at SSAB. The problem occurs when different kinds of steel should be cast during the same day. Depending on which order the groups of different steel is placed different amounts of waste is produced, the goal is to minimize this waste. Earlier work has been done on this problem and resulted in a mathematical model to create feasible solutions to this problem. In practice the time it takes to create good solutions are long and the question is if it is possible to use a heuristic method to generate good solutions in a shorter amount of time. Drawing upon inspiration from metaheuristics such as Variable Neighbourhood Search, Simualted Annealing and Tabu Search multiple heuristics have been created, implemented and evaluated against the mathematical model. One of the heuristics perform better than the mathematical model does in 10 minutes. The result from the mathematical model after 60 minutes is slightly better than the heuristic, but the results are similar. With regards to running time the heuristic takes considerably less time than 10 minutes.

Orientador: Edmea Cassia Baptista / Resumo: O problema de Despacho Econômico com Ponto de Válvula é um importante problema relacionado aos Sistemas Elétricos de Potência, que pode ser formulado como um problema de otimização não linear, não convexo e não diferenciável, o que dificulta sua resolução através de métodos exatos. Pode-se observar na literatura que diversos métodos heurísticos são propostos para a resolução do mesmo, os quais são eficientes e com um baixo custo computacional. Uma das desvantagens desses métodos é o tamanho do espaço de busca para realizer tais testes. Pesquisas realizadas apontam que, na grande maioria das vezes, os pontos ótimos para o problema de Despacho Econômico com Ponto de Válvula se encontram em pontos nos quais a função modular, presente na formulação do problema, possui valor nulo, ou estão na região destes e tais pontos são denominados de Pontos Singulares. Neste trabalho, com o bjetivo de propor um método heurístico com espaço de busca reduzido, é proposto um método de Busca Tabu direcionada a Pontos Singulares, o qual utiliza o método de Busta Tabu para percorrer os pontos nos quais a função modular se anula. O método se mostra eficiente para problemas de DEPV de 3, 13 e 40 geradores, com valores próximos aos valores ótimos obtidos por métodos determinísticos e com baixo custo computacional. / Abstract: The problem of Economic Load Dispatch with Valve Point (EDVP) is an important problem related to Electric Power Systems, that can be formulated as a non-linear, non-convex and non-differentiable optimization problem, that difficults resolution through deterministic methods. We can observe in the literature that many heuristic methods are proposed for the resolution of the same, being efficient with a low computational cost. One of the advantages of this methods is the size of the search space necessary to perform the tests. Researches points out that, in most cases, the optimal points for the Economic Load Dispatch with Valve Point problem are at points where the modular function present in the problem formulation has zero value, or in the region thereof, these points are called Singular Points. In this work is proposed, with the objective to propose a heuristic method with the search space reducted, a Tabu Search Directed to Singular Point Search, which uses he tatbu search method to the points in which the modular function cancels out. The method is efficient for resolution of Economic Load Dispatch with Valve Point problems of 3, 13 and 40 generators unities, with values close to optimal obtained by deterministic methods values and low computational cost. / Mestre

Snow removal is an important optimization problem in countries with snowfall. Bus stops can only be cleared after the adjacent street is cleared. The problem of optimizing snow removal for bus stops in an urban area is a special case of the Travelling Salesman Problem with Time Windows, where each stop only can be cleared after a certain time has passed. The solver Gurobi is used to solve the mathematical model of this problem to optimality. A local search and a tabu search is implemented. The results of the mathematical model are compared to the results of the implemented tabu search method. The results show that if a solution needs to be produced quickly, the tabu search provides better solutions than Gurobi. / Snöröjning är ett viktigt optimeringsproblem i länder med snöfall. Busshållplatsen kan bara röjas efter att den angränsande vägen är röjd. Problemet att optimera snöröjning av busshållplatser i en stad är ett Handelsresandeproblem med tidsfönster, där varje hållplats bara kan röjas efter att en tid har gått. I arbetet har vi implementerat en tabusökning för att hitta snabbt hitta bra tillåtna lösningar till problemet. För att utvärdera prestandan hos tabusökningen har vi också implementerat en matematisk modell och använt Gurobi som lösare. Resultaten visar att tabusökningen är snabbast på att hitta tillåtna lösningar av god kvalité.

The University Course Timetabling Problem (UCTP) is the problem of assigning locations (lecture halls, computer rooms) and time slots (time and date) to a set of events (lectures, labs) while satisfying a number of constraints such as avoiding double-bookings. Many variants of problem formulations exist, and most realistic variants are thought to be NP-hard. A recent trend in solving hard scheduling problems lies in the application of hybrid metaheuristics, where improvements are often found by hybridizing a population-based approach with some form of local search. In this paper, an implementation of a Genetic Algorithm (GA) that solves the UCTP is hybridized with local search in the form of Tabu Search (TS). The results show significant improvements to the performance and scalability over the non-hybridized GA. Two application strategies for the TS are investigated. The first strategy performs a switch-over from the GA to the TS, while the second interleaves the two algorithms. The effectiveness of each application strategy is seen to depend on the characteristics of the individual algorithms. / Schemaläggningsproblemet UCTP (University Course Timetabling Problem) består av problemet att tilldela platser (föreläsningssalar, laborationssalar) och tidpunkter (datum och klockslag) till en mängd tillställningar (föreläsningar, laborationer) under kravet att upprätthålla en mängd restriktioner, exempelvis att undvika dubbelbokningar. Det finns många varianter av problemformuleringen och de flesta realistiska formuleringer anses ge upphov till NP-svåra optimeringsproblem. En förhållandevis ny trend för lösningsmodeller till svåra schemaläggningsproblem ligger i tillämpningen av hybrida metaheuristiker, där förbättringar ofta ses när populationsbaserade algoritmer kombineras med någon typ av lokalsökning. I denna rapport undersöks en UCTP-lösning baserad på en Genetisk Algoritm (GA) som hybridiseratsmed en lokalsökning i form av en Tabusökning (TS). Resultaten visar på signifikanta förbättringar i prestanda och skalbarhet jämfört med den icke-hybridiserade GA:n. Två appliceringsstrategier för TS undersöks. Den första strategin utgörs av att byta algoritm från GA till TS, medan den andra utgörs av att sammanfläta de två algoritmerna. Appliceringsstrategiernas effektivitet ses bero av de individuella algoritmernas egenskaper.

In this thesis, our focus is on low-complexity, high-performance detection algorithms for multi-antenna communication receivers. A key contribution in this thesis is the demonstration that efficient algorithms from metaheuristics and machine learning can be gainfully adapted for signal detection in multi- antenna communication receivers. We first investigate a popular metaheuristic known as the reactive tabu search (RTS), a combinatorial optimization technique, to decode the transmitted signals in large-dimensional communication systems. A basic version of the RTS algorithm is shown to achieve near-optimal performance for 4-QAM in large dimensions. We then propose a method to obtain a lower bound on the BER performance of the optimal detector. This lower bound is tight at moderate to high SNRs and is useful in situations where the performance of optimal detector is needed for comparison, but cannot be obtained due to very high computational complexity. To improve the performance of the basic RTS algorithm for higher-order modulations, we propose variants of the basic RTS algorithm using layering and multiple explorations. These variants are shown to achieve near-optimal performance in higher-order QAM as well. Next, we propose a new receiver called linear regression of minimum mean square error (MMSE) residual receiver (referred to as LRR receiver). The proposed LRR receiver improves the MMSE receiver by learning a linear regression model for the error of the MMSE receiver. The LRR receiver uses pilot data to estimate the channel, and then uses locally generated training data (not transmitted over the channel) to find the linear regression parameters. The LRR receiver is suitable for applications where the channel remains constant for a long period (slow-fading channels) and performs well. Finally, we propose a receiver that uses a committee of linear receivers, whose parameters are estimated from training data using a variant of the AdaBoost algorithm, a celebrated supervised classification algorithm in ma- chine learning. We call our receiver boosted MMSE (B-MMSE) receiver. We demonstrate that the performance and complexity of the proposed B-MMSE receiver are quite attractive for multi-antenna communication receivers.

Thesis (M.Com.(quantitative Management)) / Satellite Scheduling Problems (SSP) are NP-hard and constraint programming and metaheuristics solution methods yield mixed results. This study investigates a new version of the SSP, the Satellite Constellation Time-Window Optimization Problem (SCoTWOP), involving commercial satellite constellations that provide frequent earth coverage. The SCoTWOP is related to the dual of the Vehicle Routing Problem with Multiple Timewindows, suggesting binary solution vectors representing an activation of time-windows. This representation fitted well with the MatLab® Genetic Algorithm and Direct Search Toolbox subsequently used to experiment with genetic algorithms, tabu search, and simulated annealing as SCoTWOP solution methods. The genetic algorithm was most successful and in some instances activated all 250 imaging time-windows, a number that is typical for a constellation of six satellites. / Quantitative Management