Técnicas inteligentes hídridas para o controle de sistemas não lineares

A neuro-fuzzy system consists of two or more control techniques in only one structure. The main characteristic of this structure is joining one or more good aspects from each technique to make a hybrid controller. This controller can be based in Fuzzy systems, artificial Neural Networks, Genetics Algorithms or rein forced learning techniques. Neuro-fuzzy systems have been shown as a promising technique in industrial applications. Two models of neuro-fuzzy systems were developed, an ANFIS model and a NEFCON model. Both models were applied to control a ball and beam system and they had their results and needed changes commented. Choose of inputs to controllers and the algorithms used to learning, among other information about the hybrid systems, were commented. The results show the changes in structure after learning and the conditions to use each one controller based on theirs characteristics

Um método para localização facial Invariante a rotação

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior

Separação cega de fontes lineares e não lineares usando algoritmo genético, redes neurais artificiais RBF e negentropia de Rényi como medida de independência

Conventional methods to solve the problem of blind source separation nonlinear, in general, using series of restrictions to obtain the solution, often leading to an imperfect separation of the original sources and high computational cost. In this paper, we propose an alternative measure of independence based on information theory and uses the tools of artificial intelligence to solve problems of blind source separation linear and nonlinear later. In the linear model applies genetic algorithms and Rényi of negentropy as a measure of independence to find a separation matrix from linear mixtures of signals using linear form of waves, audio and images. A comparison with two types of algorithms for Independent Component Analysis widespread in the literature. Subsequently, we use the same measure of independence, as the cost function in the genetic algorithm to recover source signals were mixed by nonlinear functions from an artificial neural network of radial base type. Genetic algorithms are powerful tools for global search, and therefore well suited for use in problems of blind source separation. Tests and analysis are through computer simulations

Odometria visual baseada em técnicas de structure from motion

Visual Odometry is the process that estimates camera position and orientation based solely on images and in features (projections of visual landmarks present in the scene) extraced from them. With the increasing advance of Computer Vision algorithms and computer processing power, the subarea known as Structure from Motion (SFM) started to supply mathematical tools composing localization systems for robotics and Augmented Reality applications, in contrast with its initial purpose of being used in inherently offline solutions aiming 3D reconstruction and image based modelling. In that way, this work proposes a pipeline to obtain relative position featuring a previously calibrated camera as positional sensor and based entirely on models and algorithms from SFM. Techniques usually applied in camera localization systems such as Kalman filters and particle filters are not used, making unnecessary additional information like probabilistic models for camera state transition. Experiments assessing both 3D reconstruction quality and camera position estimated by the system were performed, in which image sequences captured in reallistic scenarios were processed and compared to localization data gathered from a mobile robotic platform

Utilização de sistemas dedicados a protocolos de rede aplicados à eficiência energética da iluminação pública

The public illumination system of Natal/RN city presents some recurring problems in the aspect of monitoring, since currently is not possible to detect in real time the light bulbs which are on throughout the day, or those which are off or burned out, at night. These factors depreciate the efficiency of the services provided, as well as, the use of energetic resources, because there is energetic waste and, consequently, financial resources that could be applied at the own public system illumination. The purpose of the work is create a prototype in substitution to the currently photoelectric relays used at public illumination, that have the same function, as well others: turn on or off the light bulbs remotely (control flexibility by the use of specifics algorithms supervisory), checking the light bulbs status (on or off) and wireless communication with the system through the ZigBee® protocol. The development steps of this product and the tests carried out are related as a way to validate and justify its use at the public illumination

Implementação de rotinas computacionais básicas para cabeça estéreo robótica

O Laboratório de Sistemas Inteligentes do Departamento de Engenharia de Computação e Automação da Universidade Federal do Rio Grande do Norte - UFRN -tem como um de seus projetos de pesquisa -Robosense -a construção de uma plataforma robótica móvel. Trata-se de um robô provido de duas rodas, acionadas de forma diferencial, dois braços, com 5 graus de liberdade cada, um cinturão de sonares e uma cabeça estéreo. Como objetivo principal do projeto Robosense, o robô deverá ser capaz de navegar por todo o prédio do LECA, desviando de obstáculos. O sistema de navegação do robô, responsável pela geração e seguimento de rotas, atuará em malha fechada. Ou seja, sensores serão utilizados pelo sistema com o intuito de informar ao robô a sua pose atual, incluindo localização e a configuração de seus recursos. Encoders (sensores especiais de rotação) foram instalados nas rodas, bem como em todos os motores dos dois braços da cabeça estéreo. Sensores de fim-de-curso foram instalados em todas as juntas da cabeça estéreo para que seja possível sua pré-calibração. Sonares e câmeras também farão parte do grupo de sensores utilizados no projeto. O robô contará com uma plataforma composta por, a princípio, dois computadores ligados a um barramento único para uma operação em tempo real, em paralelo. Um deles será responsável pela parte de controle dos braços e de sua navegação, tomando como base as informações recebidas dos sensores das rodas e dos próximos objetivos do robô. O outro computador processará todas as informações referentes à cabeça estéreo do robô, como as imagens recebidas das câmeras. A utilização de técnicas de imageamento estéreo torna-se necessária, pois a informação de uma única imagem não determina unicamente a posição de um dado ponto correspondente no mundo. Podemos então, através da utilização de duas ou mais câmeras, recuperar a informação de profundidade da cena. A cabeça estéreo proposta nada mais é que um artefato físico que deve dar suporte a duas câmeras de vídeo, movimentá-las seguindo requisições de programas (softwares) apropriados e ser capaz de fornecer sua pose atual. Fatores como velocidade angular de movimentação das câmeras, precisão espacial e acurácia são determinantes para o eficiente resultado dos algoritmos que nesses valores se baseiam

Modelos de dispositivos de microondas e ópticos através de redes neurais artificiais de alimentação direta

This dissertation contributes for the development of methodologies through feed forward artificial neural networks for microwave and optical devices modeling. A bibliographical revision on the applications of neuro-computational techniques in the areas of microwave/optical engineering was carried through. Characteristics of networks MLP, RBF and SFNN, as well as the strategies of supervised learning had been presented. Adjustment expressions of the networks free parameters above cited had been deduced from the gradient method. Conventional method EM-ANN was applied in the modeling of microwave passive devices and optical amplifiers. For this, they had been proposals modular configurations based in networks SFNN and RBF/MLP objectifying a bigger capacity of models generalization. As for the training of the used networks, the Rprop algorithm was applied. All the algorithms used in the attainment of the models of this dissertation had been implemented in Matlab

Implementação de um sistema embarcado aplicado ao controle de pressão arterial

Hypertension is a dangerous disease that can cause serious harm to a patient health. In some situations the necessity to control this pressure is even greater, as in surgical procedures and post-surgical patients. To decrease the chances of a complication, it is necessary to reduce blood pressure as soon as possible. Continuous infusion of vasodilators drugs, such as sodium nitroprusside (SNP), rapidly decreased blood pressure in most patients, avoiding major problems. Maintaining the desired blood pressure requires constant monitoring of arterial blood pressure and frequently adjusting the drug infusion rate. Manual control of arterial blood pressure by clinical personnel is very demanding, time consuming and, as a result, sometimes of poor quality. Thus, the aim of this work is the design and implementation of a database of tuned controllers based on patients models, in order to find a suitable PID to be embedded in a Programmable Integrated Circuit (PIC), which has a smaller cost, smaller size and lower power consumption. For best results in controlling the blood pressure and choosing the adequate controller, tuning algorithms, system identification techniques and Smith predictor are used. This work also introduces a monitoring system to assist in detecting anomalies and optimize the process of patient care.

Uma contribuição à solução do problema dos k-servos usando aprendizagem por reforço

Neste trabalho é proposto um novo algoritmo online para o resolver o Problema dos k-Servos (PKS). O desempenho desta solução é comparado com o de outros algoritmos existentes na literatura, a saber, os algoritmos Harmonic e Work Function, que mostraram ser competitivos, tornando-os parâmetros de comparação significativos. Um algoritmo que apresente desempenho eficiente em relação aos mesmos tende a ser competitivo também, devendo, obviamente, se provar o referido fato. Tal prova, entretanto, foge aos objetivos do presente trabalho. O algoritmo apresentado para a solução do PKS é baseado em técnicas de aprendizagem por reforço. Para tanto, o problema foi modelado como um processo de decisão em múltiplas etapas, ao qual é aplicado o algoritmo Q-Learning, um dos métodos de solução mais populares para o estabelecimento de políticas ótimas neste tipo de problema de decisão. Entretanto, deve-se observar que a dimensão da estrutura de armazenamento utilizada pela aprendizagem por reforço para se obter a política ótima cresce em função do número de estados e de ações, que por sua vez é proporcional ao número n de nós e k de servos. Ao se analisar esse crescimento (matematicamente, ) percebe-se que o mesmo ocorre de maneira exponencial, limitando a aplicação do método a problemas de menor porte, onde o número de nós e de servos é reduzido. Este problema, denominado maldição da dimensionalidade, foi introduzido por Belmann e implica na impossibilidade de execução de um algoritmo para certas instâncias de um problema pelo esgotamento de recursos computacionais para obtenção de sua saída. De modo a evitar que a solução proposta, baseada exclusivamente na aprendizagem por reforço, seja restrita a aplicações de menor porte, propõe-se uma solução alternativa para problemas mais realistas, que envolvam um número maior de nós e de servos. Esta solução alternativa é hierarquizada e utiliza dois métodos de solução do PKS: a aprendizagem por reforço, aplicada a um número reduzido de nós obtidos a partir de um processo de agregação, e um método guloso, aplicado aos subconjuntos de nós resultantes do processo de agregação, onde o critério de escolha do agendamento dos servos é baseado na menor distância ao local de demanda

Redes neurais artificiais no ambiente de redes industriais foundation fieldbus usando blocos funcionais padrões

Industrial automation networks is in focus and is gradually replacing older architectures of systems used in automation world. Among existing automation networks, most prominent standard is the Foundation Fieldbus (FF). This particular standard was chosen for the development of this work thanks to its complete application layer specification and its user interface, organized as function blocks and that allows interoperability among different vendors' devices. Nowadays, one of most seeked solutions on industrial automation are the indirect measurements, that consist in infering a value from measures of other sensors. This can be made through implementation of the so-called software sensors. One of the most used tools in this project and in sensor implementation are artificial neural networks. The absence of a standard solution to implement neural networks in FF environment makes impossible the development of a field-indirect-measurement project, besides other projects involving neural networks, unless a closed proprietary solution is used, which dos not guarantee interoperability among network devices, specially if those are from different vendors. In order to keep the interoperability, this work's goal is develop a solution that implements artificial neural networks in Foundation Fieldbus industrial network environment, based on standard function blocks. Along the work, some results of the solution's implementation are also presented

Estudo do efeito de dipolos em arranjos de antenas adaptativas de sistemas de comunicação DS-CDMA

This master dissertation introduces a study about some aspects that determine the aplication of adaptative arrays in DS-CDMA cellular systems. Some basics concepts and your evolution in the time about celular systems was detailed here, meanly the CDMA tecnique, specialy about spread-codes and funtionaly principies. Since this, the mobile radio enviroment, with your own caracteristcs, and the basics concepts about adaptive arrays, as powerfull spacial filter was aborded. Some adaptative algorithms was introduced too, these are integrants of the signals processing, and are answerable for weights update that influency directly in the radiation pattern of array. This study is based in a numerical analysis of adaptative array system behaviors related to the used antenna and array geometry types. All the simulations was done by Mathematica 4.0 software. The results for weights convergency, square mean error, gain, array pattern and supression capacity based the analisis made here, using RLS (supervisioned) and LSDRMTA (blind) algorithms.

Sistema inteligente para determinação das direções de chegada de múltiplos sinais em arranjos de antenas

This dissertation presents a new proposal for the Direction of Arrival (DOA) detection problem for more than one signal inciding simultaneously on an antennas array with linear or planar geometry by using intelligent algorithms. The DOA estimator is developed by using techniques of Conventional Beam-forming (CBF), Blind Source Separation (BSS), and the neural estimator MRBF (Modular Structure of Radial Basis Functions). The developed MRBF estimator has its capacity extended due to the interaction with the BSS technique. The BSS makes an estimation of the steering vectors of the multiple plane waves that reach the array in the same frequency, that means, obtains to separate mixed signals without information a priori. The technique developed in this work makes possible to identify the multiple sources directions and to identify and to exclude interference sources

Simulador computacional do comportamento em regime de poços de petróleo equipados com gas-lift contínuo

Amongst the results of the AutPoc Project - Automation of Wells, established between UFRN and Petrobras with the support of the CNPq, FINEP, CTPETRO, FUNPEC, was developed a simulator for equipped wells of oil with the method of rise for continuous gas-lift. The gas-lift is a method of rise sufficiently used in production offshore (sea production), and its basic concept is to inject gas in the deep one of the producing well of oil transform it less dense in order to facilitate its displacement since the reservoir until the surface. Based in the use of tables and equations that condense the biggest number of information on characteristics of the reservoir, the well and the valves of gas injection, it is allowed, through successive interpolations, to simulate representative curves of the physical behavior of the existing characteristic variable. With a simulator that approaches a computer of real the physical conditions of an oil well is possible to analyze peculiar behaviors with very bigger speeds, since the constants of time of the system in question well are raised e, moreover, to optimize costs with assays in field. The simulator presents great versatility, with prominance the analysis of the influence of parameters, as the static pressure, relation gas-liquid, pressure in the head of the well, BSW (Relation Basic Sediments and Water) in curves of request in deep of the well and the attainment of the curve of performance of the well where it can be simulated rules of control and otimization. In moving the rules of control, the simulator allows the use in two ways of simulation: the application of the control saw software simulated enclosed in the proper simulator, as well as the use of external controllers. This implies that the simulator can be used as tool of validation of control algorithms. Through the potentialities above cited, of course one another powerful application for the simulator appears: the didactic use of the tool. It will be possible to use it in formation courses and recycling of engineers

Estudo avaliativo de um algoritmo genético auto-organizável e multiobjetivo utilizando aprendizado de máquina para aplicações de telecomunicações

This paper presents an evaluative study about the effects of using a machine learning technique on the main features of a self-organizing and multiobjective genetic algorithm (GA). A typical GA can be seen as a search technique which is usually applied in problems involving no polynomial complexity. Originally, these algorithms were designed to create methods that seek acceptable solutions to problems where the global optimum is inaccessible or difficult to obtain. At first, the GAs considered only one evaluation function and a single objective optimization. Today, however, implementations that consider several optimization objectives simultaneously (multiobjective algorithms) are common, besides allowing the change of many components of the algorithm dynamically (self-organizing algorithms). At the same time, they are also common combinations of GAs with machine learning techniques to improve some of its characteristics of performance and use. In this work, a GA with a machine learning technique was analyzed and applied in a antenna design. We used a variant of bicubic interpolation technique, called 2D Spline, as machine learning technique to estimate the behavior of a dynamic fitness function, based on the knowledge obtained from a set of laboratory experiments. This fitness function is also called evaluation function and, it is responsible for determining the fitness degree of a candidate solution (individual), in relation to others in the same population. The algorithm can be applied in many areas, including in the field of telecommunications, as projects of antennas and frequency selective surfaces. In this particular work, the presented algorithm was developed to optimize the design of a microstrip antenna, usually used in wireless communication systems for application in Ultra-Wideband (UWB). The algorithm allowed the optimization of two variables of geometry antenna - the length (Ls) and width (Ws) a slit in the ground plane with respect to three objectives: radiated signal bandwidth, return loss and central frequency deviation. These two dimensions (Ws and Ls) are used as variables in three different interpolation functions, one Spline for each optimization objective, to compose a multiobjective and aggregate fitness function. The final result proposed by the algorithm was compared with the simulation program result and the measured result of a physical prototype of the antenna built in the laboratory. In the present study, the algorithm was analyzed with respect to their success degree in relation to four important characteristics of a self-organizing multiobjective GA: performance, flexibility, scalability and accuracy. At the end of the study, it was observed a time increase in algorithm execution in comparison to a common GA, due to the time required for the machine learning process. On the plus side, we notice a sensitive gain with respect to flexibility and accuracy of results, and a prosperous path that indicates directions to the algorithm to allow the optimization problems with "η" variables

Classificação de padrões através de um comitê de máquinas aprimorado por aprendizagem por reforço

Reinforcement learning is a machine learning technique that, although finding a large number of applications, maybe is yet to reach its full potential. One of the inadequately tested possibilities is the use of reinforcement learning in combination with other methods for the solution of pattern classification problems. It is well documented in the literature the problems that support vector machine ensembles face in terms of generalization capacity. Algorithms such as Adaboost do not deal appropriately with the imbalances that arise in those situations. Several alternatives have been proposed, with varying degrees of success. This dissertation presents a new approach to building committees of support vector machines. The presented algorithm combines Adaboost algorithm with a layer of reinforcement learning to adjust committee parameters in order to avoid that imbalances on the committee components affect the generalization performance of the final hypothesis. Comparisons were made with ensembles using and not using the reinforcement learning layer, testing benchmark data sets widely known in area of pattern classification