CAPES / Este trabalho apresenta a aplicação dos métodos de otimização por enxame de partículas e por colônia de formigas na otimização dos parâmetros de um controlador não linear de estrutura variável baseado em um controlador de variância mínima generalizada. Este controlador é composto por duas parcelas distintas: uma parcela linear e outra não linear. A parcela não linear do controlador apresenta dois parâmetros que afeta diretamente o comportamento do controlador e tais parâmetros são obtidos de maneira empírica. As metaheurísticas foram aplicadas para se obter os valores otimizados destes parâmetros. Foi considerada uma função custo que leva em consideração o erro de rastreamento e a variação da ação de controle. Um exemplo numérico do projeto deste controlador também é apresentado. O controlador otimizado foi experimentado em três plantas reais: controle de velocidade de um servomecanismo, controle de nível e controle de vazão em uma planta didática industrial. Os resultados obtidos enfatizam a melhora do desempenho do controlador com os parâmetros otimizados. Também é apresentada a comparação do desempenho deste controlador com um controlador PI. / This work presents the application of particle swarm optimization and ant colony optimization in the parameters optimization of a non-linear controller with variable structure based on generalized minimum variance control. This controller is composed of two parts: linear and non-linear. The non-linear term of the controller consists of two parameters that directly affects the control action, and are obtained by trial and error. Metaheuristic methods were applied to find out the optimized values of these parameters. The cost function used in metaheuristic methods takes account the error and the control signal. A numerical example of the design of this controller is also presented. Three practical experiments were considered: a servomechanism velocity control and two control loops in a didactic industrial plant, level and flow control. Experimental results emphasize the improvement of the system performance when the optimization methods are applied. A comparision with PI controller is shown.

Made available in DSpace on 2016-03-15T19:37:57Z (GMT). No. of bitstreams: 1 DAVILA PATRICIA FERREIRA CRUZ.pdf: 3761174 bytes, checksum: 5bdf7491a01f52fa9d31b6f66eca7c87 (MD5) Previous issue date: 2015-06-17 / With the popularization of Internet, the advancement of electronic devices and the ease of storage, the volume of data stored and available at companies has increased substantially. Therefore, it becomes necessary to use intelligent techniques to extract useful information and knowledge from these data. In this context, Data Mining has been the aim of several researches by providing a set of intelligent techniques to the exploration of large volumes of data. The present project aims to research and develop new algorithms inspired by the collective behavior of bee colonies for solving complex clustering and classification tasks. More specifically, this project proposes adaptations of an optimization algorithm inspired by the behavior of bees so that it can be applied to solve clustering problems and also for positioning centers of RBF neural networks. The proposed approaches were applied to several benchmark problems with promising results. / Com a popularização da Internet, o avanço dos dispositivos eletrônicos e a facilidade de armazenamento, o volume de dados armazenados e disponibilizados por empresas de diversos ramos tem aumentado rapidamente. Com isso, torna-se necessária a utilização de técnicas avançadas capazes de extrair desses dados informações úteis e conhecimentos que, na maioria das vezes, estão implícitos. Nesse contexto, a Mineração de Dados tem sido alvo de diversas pesquisas por prover um conjunto de técnicas inteligentes para a exploração de grandes volumes de dados. O presente projeto visa à investigação e desenvolvimento de novos algoritmos inspirados no comportamento coletivo das colônias de abelhas para aplicação em problemas complexos de classificação e agrupamentos de dados, que são importantes tarefas da Mineração de Dados. Mais especificamente, esse projeto propõe adaptações de um algoritmo de otimização inspirado no comportamento de abelhas, sua aplicação em problemas de agrupamento de dados e para o posicionamento de centros de redes neurais do tipo RBF. Os resultados experimentais em bases de dados da literatura mostraram a viabilidade e benefícios das propostas, tanto para problemas de agrupamento, quanto para problemas de classificação.

The Ant Colony Optimization and Particle Swarm Optimization are two Swarm Intelligence algorithms often utilized for optimization. Swarm Intelligence relies on agents that possess fragmented knowledge, a concept not often utilized in games. The aim of this study is to research whether there are any benefits to using these Swarm Intelligence algorithms in comparison to standard algorithms such as A* for pathfinding in a game. Games often consist of dynamic environments with mobile agents, as such all experiments were conducted with dynamic destinations. Algorithms were measured on the length of their path and the time taken to calculate that path. The algorithms were implemented with minor modifications to allow them to better function in a grid based environment. The Ant Colony Optimization was modified in regards to how pheromone was distributed in the dynamic environment to better allow the algorithm to path towards a mobile target. Whereas the Particle Swarm Optimization was given set start positions and velocity in order to increase initial search space and modifications to increase particle diversity. The results obtained from the experimentation showcased that the Swarm Intelligence algorithms were capable of performing to great results in terms of calculation speed, they were however not able to obtain the same path optimality as A*. The algorithms' implementation can be improved but show potential to be useful in games.

Techniques for algorithmic musical composition or generative music working directly with the frequencies of the sounds being played are rare today as most approaches rely on mapping of discrete states. The purpose of this work is to investigate how self organizing audio can be created in realtime based on pitch information, and to find methods that give both expressive control and some unpredictability. A series of experiments were done using SuperCollider and evaluated against criteria formulated using music theory and psychoacoustics. One approach was utilizing the missing fundamental phenomenon and pitch detection using autocorrelation. This approach generated unpredictable sounds but was too much reliant on user input to generate evolving sounds. Another approach was the Kuramoto model of synchronizing oscillators. This resulted in pleasant phasing sounds when oscillators modulating the amplitudes of audible oscillators were synchronized, and distorted sounds when the frequencies of the audible oscillators were synchronized. Lastly, swarming behavior was investigated by implementing an audio analogy of Reynolds’ Boids model. The boids model resulted in interesting independently evolving sounds. Only the boids model showed true promise as a method of algorithmic composition. Further work could be done to expand the boids model by incorporating more parameters. Kuramoto synchronization could viably be used for sound design or incorporated into the boids model.

Accumulation and aggregation of the 42-residue amyloid-[beta] (A[beta]) protein fragment, which originates from the cleavage of amyloid precursor protein by beta and gamma secretase, correlates with the pathology of Alzheimer's disease (AD). Possible therapies for AD include peptides based on the A[beta] sequence, and recently identified small molecular weight compounds designed to mimic these, that interfere with the aggregation of A[beta] and prevent its toxic effects on neuronal cells in culture. Here, we use molecular dynamics simulations to compare the mode of interaction of an active (LPFFD) and inactive (LHFFD) [beta]-sheet breaker peptide with an A[beta] fibril structure from solid state NMR studies. We found that LHFFD had a weaker interaction with the fibril than the active peptide, LPFFD, from geometric and energetic considerations, as estimated by the MM/PBSA approach. Cluster analysis and computational alanine scanning identified important ligand-fibril contacts, including a possible difference in the effect of histidine on ligand-fibril [pi]-stacking interactions, and the role of the proline residue establishing contacts that compete with those essential for maintenance of the inter-monomer [beta]-sheet structure of the fibril. Our results show that molecular dynamics simulations can be a useful way to classify the stability of docking sites. These mechanistic insights into the ability of LPFFD to reverse aggregation of toxic A[beta] will guide the redesign of lead compounds, and aid in developing realistic therapies for AD and other diseases of protein aggregation. We have also performed long explicit solvent MD simulations of unliganded amyloid fibril in three putative protonation states, in order to better understand the energetic and mechanical features of the fibril receptor. Over 100 ns MD simulations, the trajectories where fibril has Glu11 and Glu22 side-chains protonated exhibit the least deviation from the initial solid state NMR structures. Free energy calculations on these rajectories suggest that the weakest fibril interface lies in the lateral rather than transverse direction and that there is little dependence on whether the lateral interface is situated at the edge or middle of the fibril. This agrees with recent reported steered molecular dynamics calculations. Secondly, in an effort to improve the ability of atomistic simulation techniques to directly resolve protein tertiary structure from primary amino acid sequence, we explore the use of a molecular dynamics technique based on swarm intelligence, called SWARM-MD, to identify the native states of two peptides, polyalanine and AEK17, as well as Trp-cage miniprotein. We find that the presence of cooperative swarm interactions significantly enhanced the efficiency of molecular dynamics simulations in predicting native conformation. However, it also is evident that the presence of outlying simulation replicas can adversely impact correctly folded replica structures. By slowly removing the swarm potential after folding simulations, the negative effect of the swarm potential can be alleviated and better agreement with experiment obtained.

The energy consumed by heating, ventilating, and air conditioning (HVAC) systems has increased in the past two decades. Thus, improving efficiency of HVAC systems has gained more and more attentions. This concern has posed challenges for modeling and optimizing HVAC systems. The traditional methods, such as analytical and statistical methods, are usually computationally complex and involve assumptions that may not hold in practice since HVAC system is a complex, nonlinear, and dynamic system. Data-mining approach is a novel science aiming at extracting system characteristics, identifying models and recognizing patterns from large-size data set. It has proved its power in modeling complex and nonlinear systems through various effective and successful applications in industrial, business, and medical areas. Classical data-mining techniques, such as neural networks and boosting tree have been largely applied into modeling HVAC systems in literature. Evolutionary computation, including swarm intelligence, have rapidly developed in the past decades and then applied to improving the performance of HVAC systems. This research focuses on modeling, optimizing, and controlling an HVAC system. Data-mining algorithms are first utilized to extract predictive models from experimental data set at Energy Resource Station in Ankeney. Evolutionary algorithms are then employed to solve the optimization models converted from the above data-driven models. In the optimization process, two set points of the HVAC system, supply air duct static pressure set point and supply air temperature set point, are controlled aiming at improving the energy efficiency and maintaining thermal comfort. The methodology presented in this research is applicable to various industrial processes other than HVAC systems.

A mobile ad hoc network is a collection of wireless mobile devices which dynamically form a temporary network, without using any existing network infrastructure or centralised administration. Each node in the network effectively becomes a router, and forwards packets towards the packet’s destination node. Ad hoc networks are characterized by frequently changing network topology, multi-hop wireless connections and the need for dynamic, efficient routing protocols. <p.This work considers the routing problem in a network of uniquely addressable sensors. These networks are encountered in many industrial applications, where the aim is to relay information from a collection of data gathering devices deployed over an area to central points. The routing problem in such networks are characterised by: <ul> <li>The overarching requirement for low power consumption, as battery powered sensors may be required to operate for years without battery replacement;</li> <li>An emphasis on reliable communication as opposed to real-time communication, it is more important for packets to arrive reliably than to arrive quickly; and</li> <li>Very scarce processing and memory resources, as these sensors are often implemented on small low-power microprocessors.</li> </ul> This work provides overviews of routing protocols in ad hoc networks, swarm intelligence, and swarm intelligence applied to ad hoc routing. Various mechanisms that are commonly encountered in ad hoc routing are experimentally evaluated under situations as close to real-life as possible. Where possible, enhancements to the mechanisms are suggested and evaluated. Finally, a routing protocol suitable for such low-power sensor networks is defined and benchmarked in various scenarios against the Ad hoc On-Demand Distance Vector (AODV) algorithm. / Dissertation (MSc)--University of Pretoria, 2005. / Computer Science / Unrestricted

Can robots of a swarm cooperate to solve together a complex cognitive problem that none of them can solve alone? TS-Swarm is a robot swarm that autonomously sequences tasks at run time and can therefore operate even if the correct order of execution is unknown at design time. The ability to sequence tasks endows robot swarms with unprecedented autonomy and is an important step towards the uptake of swarm robotics in a range of practical applications. / Doctorat en Sciences de l'ingénieur et technologie / info:eu-repo/semantics/nonPublished

A company, Unanimous AI, has created a software platform that allows individuals to come together as a group or a human swarm to make decisions. These human swarms amplify the decision-making capabilities of both the individuals and the group. One way Unanimous AI increases the swarm’s collective decision-making capabilities is by limiting the swarm to more informed individuals on the given topic. The previous way Unanimous AI selected users to enter the swarm was improved upon by a new methodology that is detailed in this study. This new methodology implements a new type of survey that collects data that is more indicative of a user’s knowledge on the subject than the previous survey. This study also identifies better metrics for predicting each user’s performance when predicting Major League Baseball game outcomes throughout a given week. This study demonstrates that the new machine learning models and data extraction schemes are approximately 12% more accurate than the currently implemented methods at predicting user performance. Finally, this study shows how predicting a user’s performance based purely on their inputs can increase the average performance of a group by limiting the group to the top predicted performers. This study shows that by limiting the group to the top predicted performers across five different weeks of MLB predictions, the average group performance was increased up to 5.5%, making this a superior method.

The background research in this Master’s thesis is focused on swarm intelligence. Further, there are two experiments described. They are based on released publications and they study behaviour of a group of robots during a puck gathering and during a target search. The actual thesis follows a repetition of these experiments in Microsoft Robotics Developer Studio (RDS), a free robotics simulation environment. The realization of both experiments in RDS is documented in detail and the achieved results are evaluated and compared with the results described in the publications. In conclusion, the thesis summarizes basic features, advantages and disadvantages of developing in RDS, based on a personal experience.