Made available in DSpace on 2016-08-29T15:33:20Z (GMT). No. of bitstreams: 1 tese_8364_dissertacao_stefano.pdf: 2078703 bytes, checksum: b7e3e083f76858033ffbf089d0223c49 (MD5) Previous issue date: 2014-11-27 / Utilizar robos autônomos capazes de planejar o seu caminho é um desafio que atrai vários pesqui quisadores na área de navegação de robôs. Neste contexto, este trabalho tem como objetivo implementar um algoritmo PSO híbrido para o planejamento de caminhos em ambientes dinâmicos. O mundo é discretizado em forma de mapas ladrilhados e cada quadrado representa ou não um obstáculo. O algoritmo proposto possui duas fases: a primeira utiliza o algoritmo A* para encontrar uma trajetória inicial viável que o algoritmo PSO otimiza na segunda fase. O ambiente de simulação de robótica CARMEN (Carnegie Mellon Robot Navigation Toolkit) foi utilizado para realização de todos os experimentos computacionais considerando cinco mapas gerados artificialmente com obstáculos estáticos ou dinâmicos. A análise dos resultados indicou que o algoritmo PSO híbrido proposto superou em qualidade de solução o PSO convencional, para essas instâncias.

Particle swarm optimization (PSO) is a well known population-based search algorithm, originally developed by Kennedy and Eberhart in 1995. The PSO has been utilized in a variety of application domains, providing a wealth of empirical evidence for its effectiveness as an optimizer. The PSO itself has undergone many alterations subsequent to its inception, some of which are fundamental to the PSO's core behavior, others have been more application specific. The fundamental alterations to the PSO have to a large extent been a result of theoretical analysis of the PSO's particle's long term trajectory. The most obvious example, is the need for velocity clamping in the original PSO. While there were empirical fndings that suggested that each particle's velocity was increasing at a rapid rate, it was only once a solid theoretical study was performed that the reason for the velocity explosion was understood. There has been a large amount of theoretical research done on the PSO, both for the deterministic model, and more recently for the stochastic model. This thesis presents an extension to the theoretical deterministic PSO model. Under the extended model, conditions for particle convergence to a point are derived. At present all theoretical PSO research is done under the stagnation assumption, in some form or another. The analysis done under the stagnation assumption is one where the personal best and neighborhood best are assumed to be non-changing. While analysis under the stagnation assumption is very informative, it could never provide a complete description of a PSO's behavior. Furthermore, the assumption implicitly removes the notion of a social network structure from the analysis. The model used in this thesis greatly weakens the stagnation assumption, by instead assuming that each particle's personal best and neighborhood best can occupy an arbitrarily large number of unique positions. Empirical results are presented to support the theoretical fndings. / Dissertation (MSc)--University of Pretoria, 2013. / gm2014 / Computer Science / Unrestricted

No description available.

Real-world optimisation problems often are of a dynamic nature. Recently, much research has been done to apply particle swarm optimisation (PSO) to dynamic environments (DE). However, these research efforts generally focused on optimising one variation of the PSO algorithm for one type of DE. The aim of this work is to develop a more comprehensive view of PSO for DEs. This thesis studies different schemes of characterising and taxonomising DEs, performance measures used to quantify the performance of optimisation algorithms applied to DEs, various adaptations of PSO to apply PSO to DEs, and the effectiveness of these approaches on different DE types. The standard PSO algorithm has shown limitations when applied to DEs. To overcome these limitations, the standard PSO can be modi ed using personal best reevaluation, change detection and response, diversity maintenance, or swarm sub-division and parallel tracking of optima. To investigate the strengths and weaknesses of these approaches, a representative sample of algorithms, namely, the standard PSO, re-evaluating PSO, reinitialising PSO, atomic PSO (APSO), quantum swarm optimisation (QSO), multi-swarm, and self-adapting multi-swarm (SAMS), are empirically analysed. These algorithms are analysed on a range of DE test cases, and their ability to detect and track optima are evaluated using performance measures designed for DEs. The experiments show that QSO, multi-swarm and reinitialising PSO provide the best results. However, the most effective approach to use depends on the dimensionality, modality and type of the DEs, as well as on the objective of the algorithm. A number of observations are also made regarding the behaviour of the swarms, and the influence of certain control parameters of the algorithms evaluated. Copyright / Dissertation (MSc)--University of Pretoria, 2012. / Computer Science / unrestricted

Doctora en Ingeniería Eléctrica / El presente trabajo aborda el problema del análisis de estabilidad, convergencia y desempeño de los sistemas adaptables fraccionarios, utilizando el enfoque de los modelos de error, problema que no ha sido abordado ni reportado en la literatura técnica hasta la fecha. Los modelos de error fraccionarios surgen al introducir las derivadas de orden fraccionario en los esquemas adaptables clásicos, ya sea describiendo la planta a controlar o identificar, o bien en las leyes de ajuste de los parámetros. Como parte del desarrollo del trabajo, se estudiaron los cuatro modelos de error conocidos hasta el momento, pero desde el punto de vista fraccionario. En todos los casos, el primer paso fue realizar exhaustivos estudios por simulación, que permitieron tener un nivel de comprensión inicial del desempeño de estos modelos de error, en cuanto a estabilidad y convergencia de los errores. Para analizar la estabilidad de estos modelos de error, fue preciso generar resultados matemáticos generales, los que también constituyen un importante aporte de esta Tesis Doctoral. Estos resultados permitieron completar el análisis de los Modelos de Error Fraccionarios 1 y 4 en su totalidad, y para ciertos casos particulares de los Modelos de Error Fraccionarios 2 y 3. En relación a la demostración de convergencia del error de salida a cero, se obtuvieron resultados analíticos para casos particulares en el Modelo de Error Fraccionario 1, y se expusieron de manera concreta las principales dificultades que han impedido, hasta el momento, generalizar estos resultados a los demás casos. También se obtuvieron otros resultados analíticos válidos para los cuatro modelos de error, que permiten afirmar que el promedio del cuadrado de la norma del error de salida, tiene una tendencia decreciente. Esto puede resultar de utilidad en algunas aplicaciones desde el punto de vista práctico. Respecto de la convergencia del error paramétrico, se logró determinar que ella está relacionada con alguna forma de excitación persistente, particular para los sistemas adaptables fraccionarios, pero no se logró dar cabal respuesta a esta interrogante. Sin embargo, se obtuvieron resultados analíticos parciales para el caso del Modelo de Error Fraccionario 1 escalar, quedando los restantes casos como parte del trabajo futuro a desarrollar en esta línea de investigación. No obstante, se expusieron las conclusiones intuitivas al respecto, obtenidas de los estudios por simulación. Finalmente, este trabajo se complementó con el diseño, implementación y análisis de dos aplicaciones de controladores fraccionarios. El primero corresponde al control por referencia a modelo de orden fraccionario para un regulador automático de voltaje, mientras que el segundo es un compendio de tres estrategias de control fraccionario para el control de posición en un sistema de levitación magnética, conocido como Anillo de Thomson.

Doctor en Ingeniería Eléctrica / La presente Tesis Doctoral explora el problema de la determinación de funciones de Lyapunov cuadráticas comunes (CQLF, por su sigla en inglés), en el marco de los sistemas conmutados, y el problema de la identificación en línea y control adaptable, en el marco de los sistemas adaptables basados en modelos de error. Ambos en el área de los sistemas dinámicos lineales y no lineales, y son resueltos aquí bajo el enfoque de la optimización basada en una herramienta llamada Optimización por Enjambre de Partículas (PSO, por su sigla en inglés). Los problemas anteriormente mencionados son de gran importancia y trascendencia en la actualidad, pues el primero entrega los elementos para la determinación de la estabilidad de sistemas lineales conmutados, y el segundo se relaciona con el control de plantas de parámetros desconocidos. Estos dos problemas poseen soluciones parciales, tanto desde el punto de vista de la optimización como de otros enfoques. Sin embargo, las soluciones existentes poseen beneficios demostrados, pero también limitaciones marcadas, que los siguen justificando como problemas abiertos. En cuanto al problema de la determinación de CQLFs, en la presente Tesis Doctoral se desarrollan dos nuevas metodologías: i) una metodología basada en PSO para la determinación de la no-existencia de una CQLF, y ii) una metodología basada en PSO para el cálculo de una CQLF. Ambas metodologías presentan evidentes mejoras comparativas respecto de las mejores soluciones actuales, con base en indicadores de desempeño objetivos. En el ámbito de los sistemas adaptables, el principal producto de la presente Tesis Doctoral es una metodología basada en PSO para el diseño de leyes de ajuste paramétrico en sistemas adaptables de tiempo discreto, representados por modelos de error. Desde este punto de vista, la investigación se centra en las propiedades de estabilidad que presenta el uso de PSO en sistemas adaptables, además de estudiar las ventajas comparativas respecto de técnicas tradicionalmente usadas como gradiente y mínimos cuadrados.

This thesis proposes several new techniques to improve the performance of speciated particle swarms in multimodal environments. We investigate how these algorithms can become more robust and adaptive, easier to use and able to solve a wider variety of optimisation problems. We then develop a technique that uses regression to vastly improve an algorithm's convergence speed without requiring extra evaluations. Speciation techniques play an important role in particle swarms. They allow an algorithm to locate multiple optima, providing the user with a choice of solutions. Speciation also provides diversity preservation, which can be critical for dynamic optimisation. By increasing diversity and tracking multiple peaks simultaneously, speciated algorithms are better able to handle the changes inherent in dynamic environments. Speciation algorithms often require a user to specify a parameter that controls how species form. This is a major drawback since the knowledge may not be available a priori. If the parameter is incorrectly set, the algorithm's performance is likely to be highly degraded. We propose using a time-based measure to control the speciation, allowing the algorithm to define species far more adaptively, using the population's characteristics and behaviour to control membership. Two new techniques presented in this thesis, ANPSO and ESPSO, use time-based convergence measures to define species. These methods are shown to be robust while still providing highly competitive performance. Both algorithms effectively optimised all of our test functions without requiring any tuning. Speciated algorithms are ideally suited to optimising dynamic environments, however the complexity of these environments makes them far more difficult to design algorithms for. To increase an algorithm's performance it is necessary to determine in what ways it should be improved. While all performance metrics allow optimisation techniques to be compared, they cannot show how to improve an algorithm. Until now this has been done largely by trial and error. This is extremely inefficient, in the same way it is inefficient trying to improve a program's speed without profiling it first. This thesis proposes a new metric that exclusively measures convergence speed. We show that an algorithm can be profiled by correlating the performance as measured by multiple metrics. By combining these two techniques, we can obtain far better insight into how best to improve an algorithm. Using this information, we then propose a local convergence enhancement that greatly increases performance by actively estimating the location of an optimum. The enhancement uses regression to fit a surface to the peak, guiding the search by estimating the peak's true location. By incorporating this technique, the algorithm is able to use the information contained within the fitness landscape far more effectively. We show that by combining the regression with an existing speciated algorithm, we are able to vastly improve the algorithm's performance. This technique will greatly enhance the utility of PSO on problems where fitness evaluations are expensive, or that require fast reaction to change.

Finding the optimal state of reality is the main purpose of the optimization process. The best variant from many possibilities is selected, and the effectiveness of the given system increases. Optimization has been applied in many real life engineering problems as in hydrological modelling. Within the hydrological case studies, the optimization process serves to estimate the best set of model parameters, or to train model weights in artificial neural networks. Particle swarm optimization (PSO) is relatively recent optimization technique, which has only a few parameters to adjust, and is easy to implement to the selected problem. The original algorithm was modified by many authors. They focused on changing the initialization of particles in the swarm, updating the population topology, adding new parameters into the equation, or incorporating shuffling mechanism into the algorithm. The modifications of PSO algorithm improve the performance of the optimization, prevent the premature convergence, and decrease computation time. Therefore, the main aims of the presented doctoral thesis consist of proposal of a new PSO modification with its implementation in C++ programming language. More PSO variants were compared and analysed, and the best methods based on benchmark problems were applied in two hydrological case studies. The first case study focused on utilization of PSO algorithms in inverse problem related to estimation of parameters of rainfall-runoff model Bilan. In the second case study, combination of artificial neural networks with PSO methods was introduced for forecasting the Standardized precipitation evapotranspiration drought index. It was found out, that particle swarm optimization is a suitable tool for solving problems in hydrological modelling. The most effective PSO modifications are the one with adaptive version of parameter of inertia weight, which updates the velocity of particles during searching through the multidimensional space via feedback information. The shuffling mechanism and redistribution of particles into complexes, at which the PSO runs separately, also significantly improve the performance. The contribution of this doctoral thesis lies in creation of new PSO modification, which was tested on benchmark problems, and was successfully applied in two hydrological case studies. The results of this thesis also extended the utilization of PSO methods in real life engineering optimization problems. All analysed PSO algorithms are available for later use within other research projects.

Submitted by Daniella Sodre (daniella.sodre@ufpe.br) on 2015-03-09T14:34:23Z No. of bitstreams: 2 Dissertaçao Anderson Sergio.pdf: 1358589 bytes, checksum: fdd2a84a1ce8a69596fa45676bc522e4 (MD5) license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) / Made available in DSpace on 2015-03-09T14:34:23Z (GMT). No. of bitstreams: 2 Dissertaçao Anderson Sergio.pdf: 1358589 bytes, checksum: fdd2a84a1ce8a69596fa45676bc522e4 (MD5) license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Previous issue date: 2013-03-07 / Reservoir Computing (RC) é um paradigma de Redes Neurais Artificiais com aplicações importantes no mundo real. RC utiliza arquitetura similar às Redes Neurais Recorrentes para processamento temporal, com a vantagem de não necessitar treinar os pesos da camada intermediária. De uma forma geral, o conceito de RC é baseado na construção de uma rede recorrente de maneira randômica (reservoir), sem alteração dos pesos. Após essa fase, uma função de regressão linear é utilizada para treinar a saída do sistema. A transformação dinâmica não-linear oferecida pelo reservoir é suficiente para que a camada de saída consiga extrair os sinais de saída utilizando um mapeamento linear simples, fazendo com que o treinamento seja consideravelmente mais rápido. Entretanto, assim como as redes neurais convencionais, Reservoir Computing possui alguns problemas. Sua utilização pode ser computacionalmente onerosa, diversos parâmetros influenciam sua eficiência e é improvável que a geração aleatória dos pesos e o treinamento da camada de saída com uma função de regressão linear simples seja a solução ideal para generalizar os dados. O PSO é um algoritmo de otimização que possui algumas vantagens sobre outras técnicas de busca global. Ele possui implementação simples e, em alguns casos, convergência mais rápida e custo computacional menor. Esta dissertação teve o objetivo de investigar a utilização do PSO (e duas de suas extensões – EPUS-PSO e APSO) na tarefa de otimizar os parâmetros globais, arquitetura e pesos do reservoir de um RC, aplicada ao problema de previsão de séries temporais. Os resultados alcançados mostraram que a otimização de Reservoir Computing com PSO, bem como com as suas extensões selecionadas, apresentaram desempenho satisfatório para todas as bases de dados estudadas – séries temporais de benchmark e bases de dados com aplicação em energia eólica. A otimização superou o desempenho de diversos trabalhos na literatura, apresentando-se como uma solução importante para o problema de previsão de séries temporais.

In this thesis, we consider maximized power allocation of non-orthogonal multiple ac- cess (NOMA) schemes since it outperforms than orthogonal multiple access (OMA) for the high altitude platform networks (HAPNETs) both in the back- haul and access links. Secondly, we propose a cluster formation (CF) algorithm and power-bandwidth resource allocation (PB-RA) for solving the resource management of HAPNETs. We adopt the particle swarm optimization (PSO) algorithm to explore the optimal de- ployment of high altitude platforms (HAPs) and unmanned aerial vehicles (UAVs) iteratively by a given swarm size. By PSO, we provide the best deployment under a given iteration number. Besides that, numerical results show that the NOMA schemes have better performance than OMA ones concerning different network control factors like the number of BSs, HAPs, and UAVs.