Le covoiturage se présente comme une solution de transport alternative qui vient soigner l’image environnementale, économique et sociétale de la voiture personnelle. Le problème du covoiturage dynamique consiste à élaborer en temps réel des tournées de véhicules optimisés, afin de répondre au mieux aux demandes instantanées de transport.C’est dans ce cadre que s’inscrivent nos travaux où l’optimisation et le temps réel sont les maître-mots. Étant donné la complexité exponentielle du problème, nous optons pour des méthodes approximatives pour le résoudre. Nous présentons notre première contribution en proposant une métaheuristique basée sur la recherche tabou. L'algorithme utilise un système de mémoire explicite et plusieurs stratégies de recherches développées pour éviter le piégeage par des optimums locaux. Ensuite, nous introduisons notre deuxième contribution qui se présente sous la forme d’une approche évolutionnaire supportée par un codage dynamique et basée sur des opérateurs génétiques contrôlés. La complexité exponentielle du problème nous amène à dévoiler notre troisième méthodologie, en proposant une approche évolutionnaire originale dans laquelle les chromosomes sont définis comme des agents autonomes et intelligents. Grâce à un protocole de négociation puissant, les Agents Chromosomes gèrent les opérateurs génétiques et orientent la recherche afin de trouver des solutions optimales dans un temps de calcul réduit. Dans la perspective d’une meilleure combinaison entre le covoiturage et les autres modes de transport, nous concevons un système baptisé DyCOS, intégrant nos approches et applications dédiées à la résolution du problème du covoiturage dynamique. / Carpooling is presented as an alternative transport solution that comes treat environmental image, economic and societal personal car. The dynamic carpooling problem is to develop real-time optimized touring vehicles to better respond to the instantaneous transport demands.Our work belongs within this context, where optimization and real time are the key words. Given the exponential complexity of the dynamic ridematching problem, we opt for the approximate methods to solve it. We present our first contribution by proposing a metaheuristic based on the multi-criteria tabu search. The proposed algorithm employs an explicit memory system and several searching strategies developed to avoid the entrapment by local solutions. Afterward, we introduce our second contribution which is in the form of an evolutionary approach supported by a dynamic coding and based on controlled genetic operators. However, the exponential complexity of the problem leads us to consider that a simple metaheuristics is not sufficient to solve effectively the problem of dynamic ridematching. It is with this in mind that we are unveiling our third solving methodology by developing an original evolutionary approach in which chromosomes are defined as autonomous and intelligent agents. Thanks to an accurate protocol negotiation, the Chromosomes Agents can control the genetic operators and guide search for finding optimal solutions within a reasonable period of time. With the prospect of a better combination between carpooling and other modes of transport, we design a system called DyCOS, integrating our approaches and applications dedicated to solving the problem of dynamic ridesharing.

Decentralised power systems with a high portion of power generated from renewable energy sources and cogeneration units (CHP) are emerging worldwide. Optimising the energy usage of such systems is a difficult task as the stochastic fluctuations of generation from renewable sources, the coupling of electrical and thermal power generation by CHP and the time dependence of necessary storage devices require new approaches. Evolutionary algorithms are able to solve the optimisation task of the energy management. They use the principles of erroneous replication and cumulative selection that can be observed in biological processes, too. Very often recombination is included in the optimisation process. Using these quite simple principles the algorithm is able to explore difficult, large and high dimensional solution spaces. It will converge to the optimal solution in most of the cases quite fast, compared to other types of optimisation algorithms. At the example of an one dimensional replicator it is derived that the convergence speed in optimising convex functions increases by several orders of magnitude even after a few cycles compared to Monte-Carlo-simulation. For several types of equipment models are developed in this work. The cost to operate a given power system for a given time span is chosen as objective function. There is a variety of parameters (more than 15) that can be set in the algorithm. With quite extensive investigations it could be shown that the product of number of replicators and the number of calculated cycles has the most important influence on the quality of the solution but the calculation time is also proportional to this number. If there are reasonable values chosen for the remaining parameters the algorithm will find appropriate solutions in adequate time in most of the cases. Although a pure evolutionary algorithm will converge to a solution the convergence speed can be greatly enhanced by extending it to a hybrid algorithm. Grouping the replicators of the first cycle in suggestive regions of the solution space by an intelligent initialisation algorithm and repairing bad solutions by introducing a Lamarckian repair algorithm makes the optimisation converge fast to good optima. The algorithm was tested using data of several existing energy systems of different structure. To optimise the energy usage in a power system with 15 different types of units the required computation time is in the range of 15 minutes. The results of this work show that extended hybrid evolutionary algorithms are suitable for integrated optimisation of energy usage in combined local energy systems. They reach better results with the same or less effort than many other optimisation methods. The developed method of optimisation of energy usage can be applied in energy systems of small and large size and complexity as optimisation computations of energy systems on the island of Cape Clear, at FH Offenburg and in the Allgäu demonstrate.

Portfolio optimisation has a number of constraints resulting from some practical matters and regulations. The closed-form mathematical solution of portfolio optimisation problems usually cannot include these constraints. Exhaustive search to reach the exact solution can take prohibitive amount of computational time. Portfolio optimisation models are also usually impaired by the estimation error problem caused by lack of ability to predict the future accurately. A number of Multi-Objective Genetic Algorithms are proposed to solve the problem with two objectives subject to cardinality constraints, floor constraints and round-lot constraints. Fuzzy logic is incorporated into the Vector Evaluated Genetic Algorithm (VEGA) to but solutions tend to cluster around a few points. Strength Pareto Evolutionary Algorithm 2 (SPEA2) gives solutions which are evenly distributed portfolio along the effective front while MOGA is more time efficient. An Evolutionary Artificial Neural Network (EANN) is proposed. It automatically evolves the ANN's initial values and structures hidden nodes and layers. The EANN gives a better performance in stock return forecasts in comparison with those of Ordinary Least Square Estimation and of Back Propagation and Elman Recurrent ANNs. Adaptation algorithms for selecting a pair of forecasting models, which are based on fuzzy logic-like rules, are proposed to select best models given an economic scenario. Their predictive performances are better than those of the comparing forecasting models. MOGA and SPEA2 are modified to include a third objective to handle model risk and are evaluated and tested for their performances. The result shows that they perform better than those without the third objective.

This thesis addresses three major issues in data mining regarding feature subset selection in large dimensionality domains, plausible reconstruction of incomplete data in cross-sectional applications, and forecasting univariate time series. For the automated selection of an optimal subset of features in real time, we present an improved hybrid algorithm: SAGA. SAGA combines the ability to avoid being trapped in local minima of Simulated Annealing with the very high convergence rate of the crossover operator of Genetic Algorithms, the strong local search ability of greedy algorithms and the high computational efficiency of generalized regression neural networks (GRNN). For imputing missing values and forecasting univariate time series, we propose a homogeneous neural network ensemble. The proposed ensemble consists of a committee of Generalized Regression Neural Networks (GRNNs) trained on different subsets of features generated by SAGA and the predictions of base classifiers are combined by a fusion rule. This approach makes it possible to discover all important interrelations between the values of the target variable and the input features. The proposed ensemble scheme has two innovative features which make it stand out amongst ensemble learning algorithms: (1) the ensemble makeup is optimized automatically by SAGA; and (2) GRNN is used for both base classifiers and the top level combiner classifier. Because of GRNN, the proposed ensemble is a dynamic weighting scheme. This is in contrast to the existing ensemble approaches which belong to the simple voting and static weighting strategy. The basic idea of the dynamic weighting procedure is to give a higher reliability weight to those scenarios that are similar to the new ones. The simulation results demonstrate the validity of the proposed ensemble model.

The popularity of e-commerce sites and applications that use recommendations and user modeling is increased recently. The development and contest in tourism calls attention of large-scale IT companies. These companies have started to work on recommendation systems and user modeling on tourism sector. Some of the clustering methodologies, neighboring methods and machine learning algorithms are commenced to use for making predictions about tourist&rsquo / s interests while he/she is traveling around the city. Recommendation ability is the most interesting thing for a tourist guide application. Recommender systems are composed of two main approaches, collaborative and content-based filtering. Collaborative filtering algorithms look for people that have similar interests and properties, while contentbased filtering methods pay attention to sole user&rsquo / s interests and properties to make recommendations. Both of the approaches have advantages and disadvantages, for that reason sometimes these two approaches are used together. Chosen method directly affects the recommendation quality, so advantages and disadvantages of both methods will be examined carefully. Recommendation of locations or services can be seen as a classification problem. Artificial intelligent systems like neural networks, genetic algorithms, particle swarm optimization algorithms, artificial immune systems are inspired from natural life and can be used as classifier systems. Artificial immune system, inspired from human immune system, has ability to classify huge numbers of different patterns. In this paper ESGuide, a tourist guide application that uses artificial immune system is examined. ESGuide application is a client-server application that helps tourists while they are traveling around the city. ESGuide has two components: Map agent and recommender agent. Map agent helps the tourist while he/she interacts with the city map. Tourist should rate the locations and items while traveling. Due to these ratings and client-server interaction, recommender agent tries to predict user interested places and items. Tourist has a chance to state if he/she likes the recommendation or not. If the tourist does not like the recommendation, new recommendation set is created and presented to the user.

Le trafic aérien en Europe représente environ 30 000 vols quotidiens actuellement. Selon les prévisions de l’organisme Eurocontrol, ce trafic devrait croître de 70% d’ici l’année 2020 pour atteindre 50 000 vols quotidiens. L’espace aérien, découpé en zones géographiques appelées secteurs de contrôle, atteindra bientôt son niveau de saturation vis-à-vis des méthodes actuelles de planification et de contrôle. Afin d’augmenter la quantité de trafic que peut absorber le système, il est nécessaire de diminuer la charge de travail des contrôleurs aériens en les aidant dans leur tâche de séparation des avions. En se fondant sur les demandes de plans de vol des compagnies aériennes, nous proposons une méthode de planification des trajectoires en 4D permettant de présenter au contrôleur un trafic dont la plupart des conflits auront été évités en avance. Cette planification s’établit en deux étapes successives, ayant chacune un unique degré de liberté : une allocation de niveaux de vol permettant la résolution des conflits en croisière puis une allocation d’heures de décollage permettant de résoudre les conflits restants. Nous présentons des modèles pour ces deux problèmes d’optimisation fortement combinatoires, que nous résolvons en utilisant la programmation par contraintes ou les algorithmes évolutionnaires, ainsi que des techniques permettant de prendre en compte des incertitudes sur les heures de décollage ou le suivi de trajectoire. Les simulations conduites sur l’espace aérien français mènent à des situations où tous les conflits sont évités, avec des retards alloués de l’ordre d’une minute en moyenne (80 à 90 minutes pour le vol le plus retardé) et un écart par rapport à l’altitude optimale limité à un niveau de vol pour la quasi totalité des vols. La prise en compte d’incertitudes de manière statique dégrade fortement ces solutions peu robustes, mais nous proposons un modèle dynamique utilisant une fenêtre glissante susceptible de prendre en compte des incertitudes de quelques minutes avec un impact réduit sur le coût de l’allocation. / Air traffic in Europe represents about 30,000 flights each day and forecasts from Eurocontrol predict a growth of 70% by 2020 (50,000 flights per day). The airspace, made up of numerous control sectors, will soon be saturated given the current planification and control methods. In order to make the system able to cope with the predicted traffic growth, the air traffic controllers workload has to be reduced by automated systems that help them handle the aircraft separation task. Based on the traffic demand by airlines, this study proposes a new planning method for 4D trajectories that provides conflict-free traffic to the controller. This planning method consists of two successive steps, each handling a unique flight parameter : a flight level allocation phase followed by a ground holding scheme.We present constraint programming models and an evolutionary algorithm to solve these large scale combinatorial optimization problems, as well as techniques for improving the robustness of the model by handling uncertainties of takeoff times and trajectory prediction. Simulations carried out over the French airspace successfully solved all conflicts, with a mean of one minute allocated delay (80 to 90 minutes for the most delayed flight) and a discrepancy from optimal altitude of one flight level for most of the flights. Handling uncertainties with a static method leads to a dramatic increase in the cost of the previous non-robust solutions. However, we propose a dynamic model to deal with this matter, based on a sliding time horizon, which is likely to be able to cope with a few minutes of uncertainty with reasonable impact on the cost of the solutions.

Submitted by Franciele Moreira (francielemoreyra@gmail.com) on 2018-11-07T12:39:53Z No. of bitstreams: 2 Dissertação - Nicolas Kemerich de Moura - 2018.pdf: 6431239 bytes, checksum: d07613ddd5a20d424fc3e761ae809118 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2018-11-08T10:12:28Z (GMT) No. of bitstreams: 2 Dissertação - Nicolas Kemerich de Moura - 2018.pdf: 6431239 bytes, checksum: d07613ddd5a20d424fc3e761ae809118 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Made available in DSpace on 2018-11-08T10:12:28Z (GMT). No. of bitstreams: 2 Dissertação - Nicolas Kemerich de Moura - 2018.pdf: 6431239 bytes, checksum: d07613ddd5a20d424fc3e761ae809118 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2018-08-16 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / This dissertation aimed to apply several computational methods of optimization and compare their performance in the evaluation of results of physicochemical tests in power transformers. The systematic generalization of the general method of evaluation and classification – normalized doubly weighted sum –, was presented, it was improved by reducing parameters and it was applied to the physicochemical tests, which allowed the evaluation of the insulating oil of power transformers through a Global Note. The results highlighted the high adaptability of the evolutionary algorithm to this specific problem. Furthermore, the high accuracy rates obtained through optimizations explained the potential of applying the weighted sum method as a tool to aid the diagnosis of power transformers, contributing to more efficient maintenance in those devices and better monitoring of the evaluation of their operating conditions, improving the reliability of the electrical system. / Esta dissertação teve por objetivo aplicar diferentes métodos computacionais de otimização e comparar seus desempenhos na avaliação dos resultados de ensaios físico-químicos em transformadores de potência. Apresentou-se a sistematização generalizada do método geral de avaliação e classificação denominado soma duplamente ponderada normalizada, que foi aprimorado por meio da redução de parâmetros, aplicado aos ensaios físico-químicos, e posteriormente otimizado, propiciando a avaliação do óleo isolante de transformadores de potência por meio de uma Nota Global. Os resultados demonstraram a capacidade do algoritmo evolutivo se adaptar muito bem ao problema específico. Ainda, as elevadas taxas de acertos obtidas por meio de otimizações explicitaram o potencial de aplicação do método da soma ponderada como ferramenta para auxílio ao diagnóstico de transformadores de potência, contribuindo para manutenções mais eficientes nesses equipamentos e um melhor acompanhamento na avaliação das suas condições operativas, impactando no aumento da confiabilidade do sistema elétrico.

A Determinação da Estrutura tridimensional de Proteínas (DEP) a partir da sua seqüência de aminoácidos é importante para a engenharia de proteínas e o desenvolvimento de novos fármacos. Uma alternativa para este problema tem sido a aplicação de técnicas de computação evolutiva. As abordagens utilizando Algoritmos Evolutivos (AEs) tem obtido resultados relevantes, porém estão restritas a pequenas proteínas, com dezenas de aminoácidos e a algumas classes de proteínas. Este trabalho propõe a investigação de uma abordagem utilizando AEs para a predição da estrutura terciária de proteínas independentemente do seu tamanho e classe. Os resultados obtidos demonstram que apesar das dificuldades encontradas a abordagem investigada constitue-se em uma alternativa em relação aos métodos clássicos de determinação da estrutura terciária das proteínas. / Protein structure determination (DEP) from aminoacid sequences is very importante to protein engineering and development of new drugs. Evolutionary computation has been aplied to this problem with relevant results. Nevertheless, Evolutionary Algorithms (EAs) can work with only proteins with few aminoacids and some protein classes. This work proposes an approach using AEs to predict protein tertiary structure independly from their size and class. The obtained results show that, despite of the difficulties that have been found, the investigate approach is a relevant alternative to classical methods to protein structure determination.

Submitted by Luciana Ferreira (lucgeral@gmail.com) on 2017-11-13T14:25:31Z No. of bitstreams: 2 Dissertação - Victor Martin de Oliveira - 2017.pdf: 4224923 bytes, checksum: df22172ea97d67bc99001b28fa5e6c8a (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2017-11-13T14:26:03Z (GMT) No. of bitstreams: 2 Dissertação - Victor Martin de Oliveira - 2017.pdf: 4224923 bytes, checksum: df22172ea97d67bc99001b28fa5e6c8a (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Made available in DSpace on 2017-11-13T14:26:03Z (GMT). No. of bitstreams: 2 Dissertação - Victor Martin de Oliveira - 2017.pdf: 4224923 bytes, checksum: df22172ea97d67bc99001b28fa5e6c8a (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2017-10-18 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / In the context of interactive art, which the spectators become interactors as well, technological development promotes new types of interaction and relations between the art and the human. The project “C³ – Cubos Interativos” (C³ project) rises in this context, created by Media Lab -- UFG with the philosophy of interpersonal and interactive relations, using art and technology. The project consists of three real cubes, which can be handled by users and produce feedback through light and sound effects. The users may communicate with one another and interact with the cubes in order to discover their behaviors and the possible reactions to the interactive activities. However, the cubes behaviors are created manually through the codification of a state machine, being a complex and time consuming task. On the other hand, the Interactive Evolutionary Computation (IEC) is an area of research that can be applied to the composition of artistic elements by using evolutionary algorithms and human interaction. One down point of the IEC is the human fatigue, what makes prohibitive the processing of many evolutionary cycles. Some techniques can be applied to avoid this problem, for example, the use of surrogate functions. This work aims to unite aspects of interactive art and interactive evolutionary computation, with the objective of providing a new way of creating behaviors that represents interesting and pleasant compositions to the C³ cubes. To achieve this goal, we propose the evolution of the C³ cubes state machines using IEC assisted by a surrogate function. A simulation environment for the C³ project was developed, in which the users can interact with virtual cubes and evaluate their behaviors, guiding the evolutionary approach. An experiment with the approach involving a group of users from UFG resulted in more complex and interesting C³ projects. / No contexto de arte interativa, em que o espectador se torna também um interator, avanços tecnológicos proporcionam novos tipos de interações e relações entre a arte e o ser humano. O projeto “C³ – Cubos Interativos” (projeto C³) surge neste contexto, criado no Media Lab -- UFG com a filosofia de relação interpessoal e interativa utilizando-se da arte e da tecnologia. Ele consiste de três cubos reais, os quais podem ser manipulados por usuários e que produzem um feedback através de efeitos luminosos e sonoros. Os usuários interagem entre si e com os cubos, a fim de descobrir seus comportamentos e as possíveis reações às atividades interativas. No entanto, a programação de comportamentos para os cubos é realizada manualmente através da codificação de uma máquina de estados, o que requer tempo e é uma tarefa complexa. Por outro lado, a computação evolutiva interativa (CEI) é uma área de pesquisa que pode ser empregada para composição de elementos artísticos pela utilização de algoritmos evolutivos e da interação humana. Uma desvantagem desta abordagem é a fadiga humana, impossibilitando assim a evolução de muitas gerações. Algumas técnicas podem ser utilizadas para contornar tal problema, como o uso de funções surrogate. Este trabalho tem por objetivo unir aspectos de arte interativa e computação evolutiva interativa, com o intuito de proporcionar uma nova forma de criação de comportamentos que caracterizem composições interessantes e agradáveis de forma automática, para os cubos do projeto C³. Para tanto, a abordagem proposta utiliza da CEI assistida por uma função surrogate, para a evolução das máquinas de estados presentes nos cubos C³. Também, é empregado um ambiente de simulação para o projeto C³, no qual usuários podem interagir com cubos virtuais e avaliar seus comportamentos, guiando o processo evolutivo. Um experimento foi realizado com um grupo de usuários da UFG, resultando em projetos C³ mais complexos e interessantes.

Mapas topologicamente ordenados sÃo tÃcnicas de representaÃÃo de dados baseadas em reduÃÃo de dimensionalidade com a propriedade especial de preservaÃÃo da vizinhanÃa espacial entre os protÃtipos no espaÃo dos dados e entre suas respectivas posiÃÃes no espaÃo de saÃda. Com base nesta propriedade, mapas topologicamente ordenados sÃo aplicados principalmente em agrupamento, quantizaÃÃo vetorial ou reduÃÃo de dimensionalidade e visualizaÃÃo de dados. Esta tese propÃe uma nova classificaÃÃo para os algoritmos de formaÃÃo de mapas topologicamente ordenados baseada no mecanismo de correlaÃÃo entre os espaÃos de entrada e de saÃda, e descreve um novo algoritmo, baseado em computaÃÃo evolucionÃria, denominado EvSOM, para a formaÃÃo de mapas topologicamente ordenado. As principais propriedades do novo algoritmo sÃo a sua flexibilidade para ponderaÃÃo pelo usuÃrio da importÃncia relativa das propriedades de quantizaÃÃo vetorial e de preservaÃÃo de topologia no mapa final, alÃm de boa rejeiÃÃo a outliers quando comparado ao algoritmo SOM de Kohonen. O trabalho desenvolve uma avaliaÃÃo empÃrica destas propriedades. O EvSOM Ã um algoritmo hÃbrido, neural-evolucionÃrio, biologicamente inspirado, que se utiliza de conceitos de redes neurais competitivas, computaÃÃo evolucionÃria, otimizaÃÃo e aproximaÃÃo iterativa. Para validar sua viabilidade de aplicaÃÃo, o EvSOM Ã estendido e especializado para a soluÃÃo de dois problemas bÃsicos relevantes em processamento de imagens e visÃo computacional, quais sejam, o problema de registro de imagens mÃdicas e o problema de rastreamento visual de objetos em vÃdeo. O algoritmo apresentou desempenho satisfatÃrio nas duas aplicaÃÃes. / Topologically ordered maps are data representation techniques based on dimensionality reduction with the special property of preserving the neighborhood between the data prototypes lying in the data space and their positions on to the output space. Based on this property, topologically ordered maps are applied mainly in clustering projected, vector quantization or dimensionality reduction and data visualization. This thesis proposes a new classification for the existing algorithms devoted to the formation of topologically ordered maps, which is based on the mechanism of correlation between the input and output spaces, and describes a new algorithm based on evolutionary computation, called EvSOM, for the topologically ordered maps formation. The main properties of the new algorithm are its flexibility for consideration by the user of the relative importance of the properties of vector quantization and topology preservation of the final map, and good outliers rejection when compared to the Kohonen SOM algorithm. The work provides an empirical evaluation of these properties. The EvSOM is a hybrid , neural-evolutionary, biologically inspired algorithm, which uses concepts of competitive neural networks, evolutionary computing, optimization and iterative approximation approximation. To validate its application feasibility, EvSOM is extended and specialized to solve two relevant basic problems in image processing and computer vision, namely, the medical image registration problem and the visual tracking of objects in video problem. The algorithm exhibits satisfactory performance in both aplications.