The task of this thesis was focused on comparison selected evolutionary algorithms for their success and computing needs. The paper discussed the basic principles and concepts of evolutionary algorithms used for optimization problems. Author programmed selected evolutionary algorithms and subsequently tasted on various test functions with exactly the given input conditions. Finally the algorithms were compared and evaluated the results obtained for different settings.

Multiobjective optimization system of exhaust manifold shapes including initial design has been developed. Space of possible solutions is explored by an evolutionary algorithm. Evaluation of exhaust drains shape comes  from drains length and sum of arc angles. Drains mustn't interfere in surrounding parts. System is tested on set of input data originated from practice. Further, performance of proposed evolutionary algorithm is evaluated.

The diploma thesis deals with the evolutionary algorithms and their application in the area of digital circuit design. In the first part, general principles of evolutionary algorithms are introduced. This part includes also the introduction of genetic algorithms and genetic programming. The next chapter describes the cartesian genetic programming and its modifications like embedded, self-modifying or multi-chromosome cartessian genetic programming. Essential part of this work consists of the design and implementation of a modularization technique for evolution circuit design. The proposed approach is evaluated using a set of standard benchmark circuits.

This thesis deals with an evolutionary solving of the Rubik's cube. The worldwide known puzzle has been for several decades not only a toy for children and adults, but also almost a lifestyle for crowds of fans and definitely a big challenge in the world of computation, where scientists seek to find an effective automated solution. The potential for its solution could also be borne by evolutionary algorithms. The author of this thesis has developed an application employing, apart from genetic algorithms, also many advanced technics, such as linear genetic programming or local search. The goal of this special technics is to make the evolutionary process more effective. There have also been made tests of the crossover, the population size and the mutation probability influence. All the tests have been statistically evaluated.

Conceptual design optimization (CDO) is a technique proposed for the structured evaluation of different design concepts. Design grammars provide a flexible modular modelling architecture. The model is generated by a compiler based on predefined components and rules. The rules describe the composition of the model; thus, different models can be optimized by the CDO in one run. This allows considering a mission design including the mission analysis and the system design. The combination of a CDO approach with a model based on design grammars is shown for the concept study of a near-Earth asteroid mission. The mission objective is to investigate two asteroids of different kinds. The CDO reveals that a mission concept using two identical spacecrafts flying to one target each is better than a mission concept with one spacecraft flying to two asteroids consecutively.

In this work two novel frameworks, one using evolutionary algorithms and another using Reinforcement Learning for training Spiking Neural Networks are proposed and analyzed. A novel multi-agent evolutionary robotics (ER) based framework, inspired by competitive evolutionary environments in nature, is demonstrated for training Spiking Neural Networks (SNN). The weights of a population of SNNs along with morphological parameters of bots they control in the ER environment are treated as phenotypes. Rules of the framework select certain bots and their SNNs for reproduction and others for elimination based on their efficacy in capturing food in a competitive environment. While the bots and their SNNs are given no explicit reward to survive or reproduce via any loss function, these drives emerge implicitly as they evolve to hunt food and survive within these rules. Their efficiency in capturing food as a function of generations exhibit the evolutionary signature of punctuated equilibria. Two evolutionary inheritance algorithms on the phenotypes, Mutation and Crossover with Mutation along with their variants, are demonstrated. Performances of these algorithms are compared using ensembles of 100 experiments for each algorithm. We find that one of the Crossover with Mutation variants promotes 40% faster learning in the SNN than mere Mutation with a statistically significant margin. Along with an evolutionary approach to training SNNs, we also describe a novel Reinforcement Learning(RL) based framework using the Proximal Policy Optimization to train a SNN for an image classification task. The experiments and results of the framework are then discussed highlighting future direction of the work.

The goal of this thesis was to explore the possibilities of using evolutionary algorithms to design components with specific purpose. We examined the process of designing an optimal shape of reflector from a highly reflective metal sheet. The main goal of this reflector is to evenly distribute light from a light emitting diode. We created a simplified model of the environment, where our component should be used. Then we used the evolutionary approach to find a suitable reflector shape for an existing device. One selected solution was manufactured and its properties measured. We also used the developed program to search for a design of an optical part for a completely new device proposal. Both tasks were accompanied by a number of problems that originated in an inaccurate task specification and general disparity between the fields of evolutionary computation and industrial components development. We provided an analysis of issues we encountered and presented solutions that can be applied to other similar tasks.

In this thesis, we present a way to model uncertainty when optimizing the UniversityTimetabling Problem. It is an NP-hard, combinatorial and highly constrained problem.In this thesis, we first propose a standardized model based on the data from MalmöUniversity. Then, we propose our extended model, which, during the creation of the solution, accounts for the probability of unexpected events to occur and changes the solution accordingly. To implement our model, we use a Particle Swarm Optimization (PSO) algorithm.In our experiments, we find problems with the algorithm converging too early.We analyze the performance of our extended model compared to the standardized model,using a benchmark devised by us, and find that it performs well, reducing the number ofconstraint violations by 32%. We then suggest further areas of research in regards to thisuncertainty model.

An Evolutionary Algorithm was run in real-time for the procedural generation ofenemies in a third-person, wave based hack and slash and shoot 'em up game. Thealgorithm evaluates enemies as individuals based on their effectiveness at battlingthe player character. Every generation is presented as a new wave of enemieswhose properties have been adjusted according to the fitness of the last wave. Byconstantly making new enemies more adept at the task of the defeating the currentplayer, I attempt to automatically and naturally raise the difficulty as the gameprogresses. The goal is also to improve player satisfaction as a result. By analyzingthe response from players and observing the changes of the generated enemies, Idetermine whether or not this is an appropriate implementation of EvolutionaryAlgorithms. Results showed that the success of the algorithm varied substantiallybetween tests, giving a number of both failed and successful tests. I go throughsome of the individual data and draw conclusions on what specific conditions makesthe algorithm perform desirably.

Le reformage autothermique, dans lequel une oxydation air carburant permet d’initier les réactions de formation d’hydrogène à partir de carburant et d’eau, semble une voie prometteuse pour la synthèse d’hydrogène à bord de navires. Son application au diesel, carburant majoritairement utilisé dans le secteur maritime, bien que moins bien connue académiquement que celle du méthane, permet une opérabilité du vaisseau sur l’ensemble du globe. Cependant les réacteurs associés sont particulièrement sujets au dépôt de carbone, néfaste pour leur durabilité, et requièrent alors une attention toute particulière au niveau des zones de mélange lors de leur conception. Dans les cas d’écoulements fortement tridimensionnels, une approche RANS couplée à un schéma cinétique décrivant les espèces gazeuses, est le plus souvent utilisée. Ce schéma consiste alors soit en un nombre succint de réactions empiriques, au risque de se montrer peu précis sur les niveaux de polluants, ou au contraire en des schémas d’une cinquantaine d’espèces issus de la réduction automatique de schémas complets, qui restent cependant trop lourds à utiliser lors d’une phase de conception. L’objectif de la thèse est alors de proposer une méthodologie pour décrire l’impact d’une géométrie sur les niveaux de polluants compatibles avec les outils habituellement utilisés dans le milieu industriel. Ainsi, la description du couplage chimie-écoulement est réalisée par le biais des logiciels Fluent R et de la suite Chemkin R de ANSYS R . Après une analyse de la chimie du reformage autothermique du diesel, une méthode de génération de schémas globaux d’une di-zaine d’espèces à partir d’un schéma détaillé est proposée. Elle est, par la suite appliquée avec succès à l’oxydation partielle du n-dodécane. Le schéma estalors utilisé dans la première simulation réactive de reformeur auto-thermique avec injection de diesel liquide réalisée à ce jour. Malgré les difficultés de validation dûes au manque de données experimentales et aux limitations des logiciels imposés, les résultats obtenus sont encourageants. / Autothermal reformers use fuel-air oxidation to ensure production of hydrogen from fuel and water on-board. The use of diesel instead of better-known methan, permits the ships to be refuelled all around the world. These systems show strong sensitivity to carbon deposit which reduces their lifetime. Good knowledge of the fuel air mixing is thus required. Academic description of such tridimensional systems usually relies on the application of a RANS simulation coupled with gaseous chemical kinetics mechanism. These mechanisms can then consist on a few empirical reactions, or at the opposite, on quite large schemes, with more than 50 species derived automatically from big detailled schemes. The resulting description is then not enough precise, or at the opposite too computationally expensive to be used during design process. This thesis thus aims to develop an industrial compatible methodology to describe the impact of design geometry on pollutant formation. ANSYS software such as Fluent and Chemkin are then used to perform the simulation. An original method of limited size mechanism derivation from larger chemical scheme is proposed. It is succesfully applied to the generation of a partial oxidation mechanism of n-dodecane, from the results of diesel reforming chemical analysis. The resulting scheme is then applied on theliquid injection diesel autoreformer reactive simulation. Even if validation difficulties result from the lack of experimental data and limitations of the softwares, it remains the first simulation of this kind in the litterature, to our knowledge. Promising results are obtained.