Submitted by Luciana Ferreira (lucgeral@gmail.com) on 2016-10-04T12:46:23Z No. of bitstreams: 2 Dissertação - Bruno Nunes Machado - 2016.pdf: 954291 bytes, checksum: 2b4b0a80a709d8803e7d0857e9aad0dd (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2016-10-04T12:46:59Z (GMT) No. of bitstreams: 2 Dissertação - Bruno Nunes Machado - 2016.pdf: 954291 bytes, checksum: 2b4b0a80a709d8803e7d0857e9aad0dd (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Made available in DSpace on 2016-10-04T12:46:59Z (GMT). No. of bitstreams: 2 Dissertação - Bruno Nunes Machado - 2016.pdf: 954291 bytes, checksum: 2b4b0a80a709d8803e7d0857e9aad0dd (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2016-09-01 / Fundação de Amparo à Pesquisa do Estado de Goiás - FAPEG / The software testing is an important component of software development life cycle, that directly affects quality of software products. Some problems in software testing phase can not be optimized only with traditional Software Engineering techniques. It is possible to do the mathematical modelling of those problems in an attempt to optimize them through the search techniques. However, the use of optimization approaches tend to incorporate more and more activities decisions to the tester, making more complex test activity. So, in order that optimization techniques are in fact employed at the Software Test solutions, the ability to abstract the details of optimization are required. Thus, the objective of this research is to propose a framework for search-based software testing (SBST). The proposed framework works as a top-level layer over generic optimization frameworks and testing software tools, it's target is supporting software testers that are not able to use optimization frameworks during a testing activity due to short deadlines and limited resources or skills, also supporting expert or beginners users from optimization area that need or want to compare their metaheuristics with ones from literature and offered by the proposed framework. The framework was evaluated in a case study of software testing scenario. This scenario was modeled as test case selection problem in which experiments were executed with different metaheuristics and benchmarks offered by framework. The results indicate it's capability to support the SBST area with emphasis on the test cases selection. The framework was evaluated and compared with other SBST frameworks in terms of quality metrics, that indicated its extensibility and flexibility as framework. / O Teste de Software é uma parte essencial do processo de desenvolvimento de software, com impacto direto na qualidade do produto de software. Alguns problemas detectados durante a fase de teste de software não são possíveis de serem resolvidos apenas com as técnicas tradicionais da Engenharia de Software. Nestes casos é possível realizar a modelagem matemática desses problemas e tentar otimizá-los por meio das técnicas de busca. Entretanto, a utilização de abordagens de otimização tende a incorporar mais decisões e mais atividades para o testador, tornando a atividade de teste mais complexa. Assim, para que as técnicas de otimização sejam de fato empregadas no Teste de Software, soluções com a capacidade de abstrair detalhes da otimização são necessárias. Diante disso, o objetivo desta pesquisa consiste em propor um framework para apoiar o Teste de Software Baseado em Busca. O framework proposto funciona como uma camada de alto nível sobre os frameworks genéricos de otimização e as ferramentas de teste de software, apoiando testadores de software que não são capazes de utilizar os frameworks de otimização durante uma atividade de teste devido a prazos curtos e recursos ou habilidades limitadas, além de apoiar usuários iniciantes ou especialistas da área de otimização que precisam ou desejam comparar suas metaheurísticas ou heurísticas com as da literatura e as oferecidas pelo framework proposto. O framework foi avaliado em um estudo de caso no cenário de teste de software. Tal cenário foi modelo como um problema de seleção de casos de teste, em que experimentos foram executados com diferentes metaheurísticas e benchmarks oferecidos pelo framework. Os resultados indicaram a capacidade do framework em apoiar a aréa de SBST, com destaque para o problema de seleção de casos de teste. Além disso, o framework também foi avaliado e comparado com outro framework SBST em termos de métricas de qualidade, que indicaram a extensibilidade e flexibilidade do framework proposto.

Submitted by Geandra Rodrigues (geandrar@gmail.com) on 2018-03-27T14:05:45Z No. of bitstreams: 1 franciscocarlosrodriguescoelho.pdf: 4772391 bytes, checksum: e11633134429c05832808dad96be9940 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2018-03-27T14:28:31Z (GMT) No. of bitstreams: 1 franciscocarlosrodriguescoelho.pdf: 4772391 bytes, checksum: e11633134429c05832808dad96be9940 (MD5) / Made available in DSpace on 2018-03-27T14:28:31Z (GMT). No. of bitstreams: 1 franciscocarlosrodriguescoelho.pdf: 4772391 bytes, checksum: e11633134429c05832808dad96be9940 (MD5) Previous issue date: 2018-02-22 / A alocação de Geração Distribuída (GD) em sistemas de distribuição de energia elétrica consiste em determinar os barramentos para conexão destas unidades geradoras, e o montante de potência a ser injetado, visando um ou mais objetivos, que podem ser: redução das perdas de potência ativa, melhorias no perfil de tensão, minimização dos custos operacionais, maximização da geração de energia, ganhos ambientais, dentre outros. O principal objetivo considerado neste trabalho é a minimização das perdas de potência ativa, mantendo as tensões dos barramentos dentro de limites recomendados. Para alcançar este objetivo, uma metodologia de otimização é proposta, tratando separadamente os problemas de localização das unidades geradoras no sistema, e o dimensionamento destas unidades. A determinação das barras com conexão de GD é realizada através de uma nova técnica de otimização metaheurística, implementada no MATLAB, denominada Metaheurística Empírica Discreta (MED). Já o dimensionamento das unidades de GD é realizado de duas formas distintas, a depender do tipo de sistema de distribuição analisado. No caso dos sistemas cujos dados são equivalentes monofásicos, o montante de potencia é determinado por um Fluxo de Potência Ótimo implementado no software comercial LINGO. A segunda estratégia de determinação da potência despachada é empregada no caso dos testes realizados com sistemas trifásicos desbalanceados, cujo dimensionamento é feito pelo método do gradiente descendente e o cálculo do fluxo de potência é realizado pelo software OpenDSS. Os três sistemas equivalentes monofásicos utilizados são compostos por 33, 69 e 476 barras, enquanto os dois trifásicos desequilibrados possuem 34 e 123 barras. A qualidade da metodologia proposta na resolução do problema de alocação de geração distribuída é avaliada através de comparações com a literatura especializada, comparações com outras metaheurísticas e testes de robustez. Os resultados provenientes de simulações com alocação de três e quatro unidades de GD em sistemas de distribuição de energia elétrica mostram que a metodologia proposta é eficiente, sendo capaz de produzir resultados com significativas reduções nas perdas de potência ativa e perfis de tensão adequados. / The optimal Distributed Generation (DG) allocation problem consists in choosing the best locations of those distributed power plants at the distribution system, and to define its amount of power injection. The approach can be either single or multiobjective. The main objectives are: minimization of total power loss, voltage profile improvement, operational cost minimization, maximization of distributed generation capacity, environmental gains, among others. In this work, the main goal pursued is the total power loss minimization of the distribution system, keeping the buses voltages within the predetermined limits. To achieve this goal, an optimization methodology is proposed. This approach treats separately the location problem and the power dispatched by the generation units. The busbars connected to distributed generation are determined through a new metaheuristic algorithm, implemented in MATLAB, named Empirical Discrete Metaheuristic (EDM). The amount of power injection is solved by an Optimum Power Flow implemented in the commercial software LINGO, or by the Steepest Descent Method in the MATLAB environment. The first strategy to determine the DG dispatch is used on simulations with single phase equivalents systems. The second one is employed in the amount of power determination in unbalanced three phase systems, which the power flow is carried out by the open source software OpenDSS. The three single phase equivalent test systems analyzed are composed by 33, 69 and 476 buses, while the two systems with three phases have 34 and 123 buses, each. To evaluate the proposed methodology quality, comparisons to published works in the specialized literature are made. Also, robustness tests and comparisons to other well succeed metaheuristics are carried out. The results were obtained from simulations with three and four DG units in electric power distribution systems. These results consistently show that the proposed methodology is efficient, providing DGs configurations that significantly reduces the active power losses and keep the voltages at adequate levels.

Submitted by Patrícia Cerveira (pcerveira1@gmail.com) on 2017-06-13T15:56:56Z No. of bitstreams: 1 Rene_Cruz_Freire.pdf: 5170376 bytes, checksum: 8c6b6dd8986d23b53ae99ba90dd69ef5 (MD5) / Approved for entry into archive by Biblioteca da Escola de Engenharia (bee@ndc.uff.br) on 2017-06-29T16:38:47Z (GMT) No. of bitstreams: 1 Rene_Cruz_Freire.pdf: 5170376 bytes, checksum: 8c6b6dd8986d23b53ae99ba90dd69ef5 (MD5) / Made available in DSpace on 2017-06-29T16:38:47Z (GMT). No. of bitstreams: 1 Rene_Cruz_Freire.pdf: 5170376 bytes, checksum: 8c6b6dd8986d23b53ae99ba90dd69ef5 (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Um dos temas de alta relevância para a sociedade atual é a qualidade do suprimento de energia elétrica, que deve ser ininterrupto, seguro e econômico. Para tal, é primordial que o sistema de potência esteja preparado para um possível defeito de algum equipamento da rede, mantendo a operação dentro dos patamares seguros, evitando os blecautes e todas as suas consequências para a sociedade. Isso pode ser feito através do redespacho das unidades geradoras, de modo a encontrar um ponto de operação que concilie segurança e economicidade, dois objetivos conflitantes, enquanto busca se afastar o mínimo possível do ponto de operação previamente estabelecido, via planejamento eletroenergético, para o sistema de potência em questão. Trata-se de uma abordagem multiobjetiva do Fluxo de Potência Ótimo com Restrições de Segurança (FPORS) que pode ser solucionada com uma abordagem de Computação Evolucionária (CE) com viés multiobjetivo. Neste trabalho, foram implementadas e comparadas duas meta-heurísticas evolutivas multiobjetivo: Nondominated Sorting Genetic Algorithm II (NSGA-II) e o Multi-objective Evolutionary Particle Swarm Optimization (MOEPSO). Os resultados dessas heurísticas também foram comparados com a abordagem mono-objetivo do mesmo problema. Os algoritmos foram implementados no MATLAB® e testados em um sistema-teste que simula as condições do Sistema Interligado Nacional (SIN). As heurísticas multiobjetivo foram comparadas através da metodologia de análise da Fronteira de Pareto (FP), onde é analisado qual método concilia melhor os objetivos de economia e segurança. Na primeira análise o NSGA-II saiu-se melhor, entretanto após a implementação de melhorias no algoritmo, o MOEPSO mostrou desempenho superior na segunda análise. Nas duas análises, o viés multiobjetivo mostrou-se superior ao mono-objetivo, na comparação através do critério de agregação de objetivos. Em relação ao tempo de simulação de cada método, o MOEPSO foi superior na primeira análise, já na segunda análise foi implementado um refinamento baseado no Fluxo de Potência Linearizado no FPORS, que baixou o tempo de simulação das duas heurísticas multiobjetivas em comparação com a primeira análise, e o MOEPSO teve o menor tempo de simulação. Na comparação com o viés mono-objetivo, apenas o NSGA-II teve tempo médio de simulação maior que o método mono-objetivo na primeira análise. Na segunda análise, todas as heurísticas multiobjetivo possuíam tempo de simulação menores que o método mono-objetivo. / One of the topics of high relevance to the today’s society is the quality of electric power supply, which must be uninterrupted, safe and economical. To this end, it is essential that the power system be prepared for a possible defect of some equipment from the network while maintaining operation within safe levels, avoiding blackouts and all its consequences for society. This can be done by redispatch of generating units, in order to find an operation point which conciliate security and economy, two conflicting objectives, while seeking to depart as little as possible of the operation point previously established in the energy planning for the power system in question. This is a multi-objective approach to Security Constrained Optimal Power Flow (SCOPF) that can be solved with an approach of Evolutionary Computation with multi-objective bias. In this work we were implemented and compared two multi-objective evolutionary meta-heuristics: Nondominated Sorting Genetic Algorithm II (NSGA-II) and Multi-objective Evolutionary Particle Swarm Optimization (MOEPSO). The results of these heuristics were also compared with mono-objective approach to the same problem. The algorithms were implemented in MATLAB® and tested in a test-case that simulates the conditions of the Brazilian Sistema Interligado Nacional (National Interconnected System). The multi-objective heuristics were compared using the analysis methodology of the Pareto Frontier, where is analyzed which method is better to conciliate the economy and security objectives. In the first analysis the NSGA-II fared better, but after the implementation of improvements in the algorithm, the MOEPSO showed superior performance in the second analisys. In both analyzes, the multi-objective bias was superior to the mono-objective bias, in the comparison through objectives aggregation criteria. Concerning the simulation time of each method, the MOEPSO was superior in the first analysis, but in the second analysis was implemented a refinement based on DC Load Flow, which lowered the simulation time of the two multi-objective heuristics compared with the first analysis, and the MOEPSO had the shortest time simulation. Compared to the mono-objective bias, only the NSGA-II had an average time simulation greater than the mono-objective method in the first analysis. In the second analysis, all multi-objectives heuristics had simulation time smaller than the mono-objective method.

Cette thèse aborde le problème de tournées de véhicules (VRP) intégrant des contraintes temporelles : fenêtres de temps (TW), synchronisation (S) et précédence (P), appliqué au secteur de soins à domicile, donnant le VRPTW-SP. Il s’agit d’établir un plan de visite journalier des soignants, aux domiciles des patients ayant besoin d’un ou plusieurs services. Tout d’abord, nous avons abordé ce problème sous angle mono-objectif. Ensuite, le cas bi-objectif est considéré. Pour la version mono-objectif, un Programme Linéaire à Variables Mixtes Entières (PLME), deux heuristiques constructives, deux procédures de recherches locales et trois métaheuristiques à base de voisinages sont proposés : une procédure de recherche constructive adaptative randomisée (GRASP), une recherche locale itérée (ILS) et une approche hybride (GRASP × ILS). Concernant le cas bi-objectif, différentes versions de métaheuristiques évolutionnaires multi-objectifs sont proposées, intégrant différentes recherches locales : l’algorithme génétique avec tri par non-dominance version 2 (NSGAII), une version généralisée de ce dernier avec démarrages multiples (MS-NSGAII) et une recherche locale itérée avec tri par non-dominance (NSILS). Ces algorithmes ont été testés et validés sur des instances adaptées de la littérature. Enfin, nous avons étendu le VRPTW-SP sur un horizon de planification, donnant le VRPTW-SP multi-période. Pour résoudre cette extension, un PLME ainsi qu’une matheuristique sont proposés / This work addresses the vehicle routing problem (VRP) including timing constraints: time windows (TW), synchronization (S) and precedence (P), applied in Home Health Care sector; giving the VRPTW-SP. This problem consists in establishing a daily caregivers planning to patients' homes asking for one or several services. We have started by considering the problem as a single objective case. Then, a bi-objective version of the problem is introduced. For solving the single-objective problem, a Mixed Integer Linear Program (MILP), two constructive heuristics, local search procedures and three local search based metaheuristics are proposed : a Greedy Randomized Adaptive Search procedure (GRASP), an Iterated Local Search (ILS) and a hybrid approach (GRASP × ILS). Regarding the bi-objective VRPTW-SP, different versions of multi-objective evolutionary algorithm, including various local research strategies are proposed: the Non-dominated Sorting Genetic Algorithm version 2 (NSGAII), a generalized version of this latter with multiple restarts (MS-NSGAII) and an Iterated Local Search combined with the Non-dominated Sorting concept (NSILS). All these algorithms have been tested and validated on appropriate instances adapted from the literature. Finally, we extended the VRPTW-SP on a multi-period planning horizon and then proposed a MILP and a matheuristic approach

Les travaux de thèse présentés s’intéressent à l’optimisation des activités d’un bloc opératoire. Ces activités concernent les interventions chirurgicales à planifier et les interventions de maintenance préventive sur les équipements dans les salles d’opération. Une solution est la synchronisation de ces activités lors de la construction du planning opératoire au niveau opératoire. Nous dissocions deux stratégies de programmation opératoire : programmation ouverte et programmation avec allocation préalable des plages horaires aux chirurgiens. Pour chacune des stratégies, nous considérons deux cas : le cas où l’heure de début d’une intervention de maintenance dans la salle est fixée, ladite intervention précédant l’affection des interventions chirurgicales dans les salles. Le second cas étant celui où l’heure de début de maintenance varie dans un intervalle entre une heure de début minimum et une heure de début maximum, avec l’intervention de maintenance placée a posteriori.Nous faisons plusieurs propositions de méthodes (exactes et approchées), y compris une méthode hybride, qui repose sur le couplage entre une métaheuristique et une heuristique. Les résultats obtenus sur des instances générées en concertation avec le monde hospitalier sont intéressants. / The presented dissertation is about the optimization of hospital systems, more precisely the optimization of the activities of an operation theatre. These activities showcase the surgical procedures to be planned and the preventive maintenance interventions on the equipment in the operating rooms. One solution is the synchronization of these activities during the construction of the operational planning at the operational level.We dissociate two operating programming strategies: Open Scheduling or Open programming and Block Scheduling or Programming with prior allocation of times to surgeons. For each strategy two cases are considered: the first case is where the time of beginning of a maintenance intervention in the room is fixed - this intervention preceding the affection of the surgical interventions in the rooms. The second case is where the maintenance start time varies in the interval between a minimum start time and a maximum start time, with the maintenance intervention placed beforehand. We make several proposition’s methods (exact and approximate), including a hybrid method, which is based on the coupling between a metaheuristic and a heuristic. The results obtained on bodies generated in consultation with the hospital’s world are interesting.

Combinatorial optimization problems are of high academical as well as practical importance. Many instances of relevant combinatorial optimization problems are, due to their dimensions, intractable for complete methods such as branch and bound. Therefore, approximate algorithms such as metaheuristics received much attention in the past 20 years. Examples of metaheuristics are simulated annealing, tabu search, and evolutionary computation. One of the most recent metaheuristics is ant colony optimization (ACO), which was developed by Prof. M. Dorigo (who is the supervisor of this thesis) and colleagues. This thesis deals with theoretical as well as practical aspects of ant colony optimization.<p><p>* A survey of metaheuristics. Chapter 1 gives an extensive overview on the nowadays most important metaheuristics. This overview points out the importance of two important concepts in metaheuristics: intensification and diversification. <p><p>* The hyper-cube framework. Chapter 2 introduces a new framework for implementing ACO algorithms. This framework brings two main benefits to ACO researchers. First, from the point of view of the theoretician: we prove that Ant System (the first ACO algorithm to be proposed in the literature) in the hyper-cube framework generates solutions whose expected quality monotonically increases with the number of algorithm iterations when applied to unconstrained problems. Second, from the point of view of the experimental researcher, we show through examples that the implementation of ACO algorithms in the hyper-cube framework increases their robustness and makes the handling of the pheromone values easier.<p><p>* Deception. In the first part of Chapter 3 we formally define the notions of first and second order deception in ant colony optimization. Hereby, first order deception corresponds to deception as defined in the field of evolutionary computation and is therefore a bias introduced by the problem (instance) to be solved. Second order deception is an ACO-specific phenomenon. It describes the observation that the quality of the solutions generated by ACO algorithms may decrease over time in certain settings. In the second part of Chapter 3 we propose different ways of avoiding second order deception.<p><p>* ACO for the KCT problem. In Chapter 4 we outline an ACO algorithm for the edge-weighted k-cardinality tree (KCT) problem. This algorithm is implemented in the hyper-cube framework and uses a pheromone model that was determined to be well-working in Chapter 3. Together with the evolutionary computation and the tabu search approaches that we develop in Chapter 4, this ACO algorithm belongs to the current state-of-the-art algorithms for the KCT problem.<p><p>* ACO for the GSS problem. Chapter 5 describes a new ACO algorithm for the group shop scheduling (GSS) problem, which is a general shop scheduling problem that includes among others the well-known job shop scheduling (JSS) and the open shop scheduling (OSS) problems. This ACO algorithm, which is implemented in the hyper-cube framework and which uses a new pheromone model that was experimentally tested in Chapter 3, is currently the best ACO algorithm for the JSS as well as the OSS problem. In particular when applied to OSS problem instances, this algorithm obtains excellent results, improving the best known solution for several OSS benchmark instances. A final contribution of this thesis is the development of a general method for the solution of combinatorial optimization problems which we refer to as Beam-ACO. This method is a hybrid between ACO and a tree search technique known as beam search. We show that Beam-ACO is currently a state-of-the-art method for the application to the existing open shop scheduling (OSS) problem instances.<p><p> / Doctorat en sciences appliquées / info:eu-repo/semantics/nonPublished

Les travaux de thèse présentés s’intéressent à l’optimisation des systèmes hospitaliers. Une solution existante est la mutualisation de ressources au sein d’un même territoire. Cela peut passer par différentes formes de coopération dont la Communauté Hospitalière de Territoire. Différents problèmes sont définis en fonction du niveau de décision : stratégique, tactique ou opérationnel ; et du niveau de modélisation : macroscopique, mesoscopique et microscopique. Des problèmes de dimensionnement, de planification et d’ordonnancement peuvent être considérés. Nous définissons notamment le problème de planification d’activités avec affectation de ressources. Plusieurs cas sont dissociés : soit les ressources humaines sont à capacité infinie, soit elles sont à capacité limitée et leur affectation sur site est une donnée, soit elles sont à capacité limitée et leur affectation sur site est une variable. Ces problèmes sont spécifiés et formalisés mathématiquement. Tous ces problèmes sont comparés à un problème de bin packing : le problème du bin packing de base pour le problème où les ressources humaines sont à capacité infinie, le problème du bin packing avec interdépendances dans les deux autres cas. Le problème du bin packing avec incompatibilités est ainsi défini. De nombreuses méthodes de résolution ont déjà été proposées pour le problème du bin packing. Nous faisons plusieurs propositions dont un couplage hiérarchique entre une heuristique et une métaheuristique. Des métaheuristiques basées individu et une métaheuristique basée population, l’optimisation par essaim particulaire, sont utilisées. Cette proposition nécessite un nouveau codage inspiré des problèmes de permutation d’ordonnancement. Cette méthode donne de très bons résultats sur les instances du problème du bin packing. Elle est simple à appliquer : elle couple des méthodes déjà connues. Grâce au couplage proposé, les nouvelles contraintes à considérer nécessitent d’être intégrées uniquement au niveau de l’heuristique. Le fonctionnement de la métaheuristique reste le même. Ainsi, notre méthode est facilement adaptable au problème de planification d’activités avec affectation de ressources. Pour les instances de grande taille, le solveur utilisé comme référence ne donne qu’un intervalle de solutions. Les résultats de notre méthode sont une fois encore très prometteurs : les solutions obtenues sont meilleures que la borne supérieure retournée par le solveur. Il est envisageable d’adapter notre méthode sur d’autres problèmes plus complexes par intégration dans l’heuristique des nouvelles contraintes à considérer. Il serait notamment intéressant de tester ces méthodes sur de réelles instances hospitalières afin d’évaluer leur portée. / The presented work is about optimization of the hospital system. An existing solution is the pooling of resources within the same territory. This may involve different forms of cooperation between several hospitals. Various problems are defined at the decision level : strategic, tactical or operational ; and at the modeling level : macroscopic, mesoscopic and microscopic. Problems of sizing, planning and scheduling may be considered. We define the problem of activities planning with resource allocation. Several cases are dissociated : either human resources are under infinite capacity, or they are under limited capacity and their assignment on a place is given, or they are under limited capacity and their assignment is a variable. These problems are specified and mathematically formalized. All thes problems are compared to a bin packing problem : the classical problem of bin packing is used for the problem where human resources are under infinite capacity, the bin packing problem with interdependencies is used in the two other cases. The bin packing problem with incompatibilities is defined. Many resolution methods have been proposed for the bin packing problem. We make several propositions including a hierarchical coupling between heuristic and metaheuristic. Single based metaheuristics and a population based metaheuristic, the particle swarm optimization, are used. This proposition requires a new encoding inspired by permutation problems. This method gives very good results to solve instances of the bin packing problem. It is easy to apply : it combines already known methods. With the proposed coupling, the new constraints to be considered need to be integrated only on the heuristic level. The running of the metaheuristic is the same. Thus, our method is easily adaptable to the problem of activities planning with resource allocation. For big instances, the solver used as a reference returns only an interval of solutions. The results of our method are once again very promising : the obtained solutions are better than the upper limit returned by the solver. It is possible to adapt our method on more complex issues through integration into the heuristic of the new constraints to consider. It would be particularly interesting to test these methods on real hospital authorities to assess their significance.

Testing of multi-threaded programs is a demanding work due to the many possible thread interleavings one should examine. The noise injection technique helps to increase the number of tested thread interleavings by noise injection to suitable program locations. This work optimizes meta-heuristics search techniques in the testing of concurrent programs by utilizing deterministic heuristic in the application of genetic algorithms in a space of legal program locations suitable for the noise injection. In this work, several novel deterministic noise injection heuristics without dependency on the random number generator are proposed in contrary to the most of currently used heuristic. The elimination of the randomness should make the search process more informed and provide better, more optimal, solutions thanks to increased stability in the results provided by novel heuristics. Finally, a benchmark of programs, used for the evaluation of novel noise injection heuristics is presented.

V disertační práci je nejprve uvedena taxonomie chyb v souběžném zpracování dat a přehled technik pro jejich dynamickou detekci. Následně jsou navrženy nové metriky pro měření synchronizace a souběžného chování programů společně s metodologií jejich odvozování. Tyto techniky se zejména uplatní v testování využívajícím techniky prohledávání prostoru a v saturačním testování. Práce dále představuje novou heuristiku vkládání šumu, jejímž cílem je maximalizace proložení instrukcí pozorovaných během testování. Tato heuristika je porovnána s již existujícími heuristikami na několika testech. Výsledky ukazují, že nová heuristika překonává ty existující v určitých případech. Nakonec práce představuje inovativní aplikaci stochastických optimalizačních algoritmů v procesu testování vícevláknových aplikací. Principem metody je hledání vhodných kombinací parametrů testů a metod vkládání šumu. Tato metoda byla prototypově implementována a otestována na množině testovacích příkladů. Výsledky ukazují, že metoda má potenciál vyznamně vylepšit testování vícevláknových programů.

[pt] Empresas de exploração de petróleo e gás offshore frequentemente precisam lidar com problemas relacionados ao uso eficiente de seus recursos. Neste trabalho, abordamos um problema de programação de navios associado à logística offshore de petróleo e gás – O Problema de Programação de Embarcações do tipo Pipe-Laying support Vessel (PLSVSP). Essas embarcações são especialmente projetadas para realizar conexões de dutos entre poços de petróleo submarinos e plataformas de produção. A conexão de dutos é a última etapa a ser executada para permitir a drenagem do óleo e iniciar a produção em um poço. No PLSVSP, o objetivo é antecipar a conclusão de poços mais produtivos. O problema pode ser visto como uma variante de um problema de programação de lotes com máquinas paralelas idênticas e tempos de configuração não antecipados por família para minimizar o total weighted completion time. Nessa analogia, embarcações são as máquinas, poços são as tarefas e lotes são as viagens executadas por PLSVs, definindo quais poços devem ser conectados a cada saída do porto. Foram desenvolvidas diversas abordagens de otimização para resolver as variantes determinística e estocástica do problema. Para a variante determinística, desenvolvemos métodos híbridos e uma metaheurística capazes de melhorar as soluções desenvolvidas por formulações MIP puras e lidar com o PLSVSP. Para a variante estocástica, foi desenvolvida uma simheurística utilizando simulação de Monte Carlo incorporada, considerando incertezas nas durações das conexões e nas datas de chegada dos oleodutos no porto. Os resultados mostram uma melhora significativa no custo das soluções quando lidam com incertezas em comparação com soluções geradas por um método determinístico. O uso da simulação em uma estrutura metaheurística mostrou-se uma abordagem promissora, capaz de lidar com o problema estocástico, com pouco esforço computacional extra necessário. / [en] Offshore oil and gas exploration companies frequently need to deal with problems related to the efficient use of their resources. In this work, we address a ship scheduling problem associated with offshore oil and gas logistics – The Pipe Laying Support Vessel Scheduling Problem (PLSVSP). These vessels are specially designed to perform pipeline connections between sub-sea oil wells and production platforms. The connections are the last step to be performed to allow the oil draining, starting production in a well. The PLSVSP objective is to anticipate the completion of the most productive wells. The problem can be seen as a variant of a batch scheduling problem with identical parallel machines and non-anticipatory family setup times to minimize the total weighted completion time. In this analogy, vessels are machines, wells are jobs, and batches are voyages executed by PLSVs, defining which wells to connect each time it leaves the port. We developed several optimization approaches to solve the deterministic and stochastic variants of the problem. For the deterministic problem, we developed hybrid methods and a metaheuristic that outperformed the pure MIP formulations, being practical to deal with the PLSVSP. A simheuristic using embedded Monte Carlo simulation was developed for the stochastic variant of the problem, considering uncertainties in the connection duration and the arrival dates of pipelines at the port. The results show a significant improvement in the solutions dealing with uncertainties compared to solutions generated by a deterministic method. The use of simulation within a metaheuristic framework proved to be a promising approach, being able to deal with the stochastic problem, with little extra computational effort required.