Submitted by Rosivalda Pereira (mrs.pereira@ufma.br) on 2017-07-03T18:32:31Z No. of bitstreams: 1 SelmoRodrigues.pdf: 1374245 bytes, checksum: 96afcfa04ba5cc18c4db55e4c92cdf23 (MD5) / Made available in DSpace on 2017-07-03T18:32:31Z (GMT). No. of bitstreams: 1 SelmoRodrigues.pdf: 1374245 bytes, checksum: 96afcfa04ba5cc18c4db55e4c92cdf23 (MD5) Previous issue date: 2017-01-26 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / This paper proposes a new methodology for modelling based on an evolving Neuro-Fuzzy Network Takagi-Sugeno (NFN-TS) for seasonal time series forecasting. The NFN-TS use the unobservable components extracted from the time series to evolve, i.e., to adapt and to adjust its structure, where the number of fuzzy rules of this network can increase or reduced according the components behavior. The method used to extract the components is a recursive version developed in this paper based on the Spectral Singular Analysis (SSA) technique. The proposed methodology has the principle divide to conquer, i.e., it divides a problem into easier subproblems, forecasting separately each component because they present dynamic behaviors that are simpler to forecast. The consequent propositions of fuzzy rules are linear state space models, where the states are the unobservable components data. When there are available observations from the time series, the training stage of NFN-TS is performed, i.e., the NFN-TS evolves its structure and adapts its parameters to carry out the mapping between the components data and the available sample of original time series. On the other hand, if this observation is not available, the network considers the forecasting stage, keeping its structure fixed and using the states of consequent fuzzy rules to feedback the components data to NFN-TS. The NFN-TS was evaluated and compared with other recent and traditional techniques for forecasting seasonal time series, obtaining competitive and advantageous results in relation to other papers. This paper also presents a case study of proposed methodology for real-time detection of anomalies based on a patient’s electrocardiogram data. / Esse trabalho propõe uma nova metodologia para modelagem baseada em uma Rede Neuro- Fuzzy Takagi-Sugeno (RNF-TS) evolutiva para a previsão de séries temporais sazonais. A RNF-TS considera as componentes não-observáveis extraídas a partir da série para evoluir, ou seja, adaptar e ajustar sua estrutura, sendo que a quantidade de regras fuzzy dessa rede pode aumentar ou ser reduzida conforme o comportamento das componentes. O método utilizado para extrair as componentes é uma versão recursiva desenvolvida nessa pesquisa baseada na técnica de Análise Espectral Singular (AES). A metodologia proposta tem como princípio dividir para conquistar, isto é, dividir um problema em subproblemas mais fáceis de lidar, realizando a previsão separadamente de cada componente já que apresentam comportamentos dinâmicos mais simples de prever. As proposições do consequente das regras fuzzy são modelos lineares no espaço de estados, sendo que os estados são os próprios dados das componentes não-observáveis. Quando há observações disponíveis da série temporal, o estágio de treinamento da RNF-TS é realizado, ou seja, a RNF-TS evolui sua estrutura e adapta seus parâmetros para realizar o mapeamento entre os dados das componentes e a amostra disponível da série temporal original. Caso contrário, se essa observação não está disponível, a rede aciona o estágio de previsão, mantendo sua estrutura fixa e usando os estados dos consequentes das regras fuzzy para realimentar os dados das componentes para a RNF-TS. A RNF-TS foi avaliada e comparada com outras técnicas recentes e tradicionais para previsão de séries temporais sazonais, obtendo resultados competitivos e vantajosos em relação a outras pesquisas. Este trabalho apresenta também um estudo de caso da metodologia proposta para detecção em tempo-real de anomalias baseada em dados de eletrocardiogramas de um paciente.

[pt] Nesta dissertação foram desenvolvidas técnicas de controle por aprendizado acelerado e Neuro-Fuzzy, aplicadas em um sistema servo-hidráulico para ensaio de fadiga. Este sistema tem o propósito de fazer ensaios em materiais para prever a resistência à fadiga dos materiais. O trabalho envolveu quatro etapas principais: levantamento bibliográfico, desenvolvimento de um controle por aprendizado acelerado, desenvolvimento de um controle por aprendizado Neuro-Fuzzy, e implementação experimental dos modelos de controle por aprendizado proposto em uma máquina de ensaios de materiais. A implementação do controle por aprendizado acelerado foi feita a partir do modelo de controle desenvolvido por Alva, com o objetivo de acelerar o processo de aprendizagem. Esta metodologia consiste em fazer um controle do tipo bang-bang, restringindo a servo-válvula a trabalhar sempre em seus limites extremos de operação, i.e., procurando mantê-la sempre completamente aberta em uma ou outra direção. Para manter a servo-válvula trabalhando em seus limites de seu funcionamento, os instantes ótimos para as reversões são obtidos pelo algoritmo de aprendizado, e armazenados em tabelas específicas para cada tipo de carregamento. Estes pontos de reversão dependem de diversos fatores, como a amplitude e carga média da solicitação, e são influenciados pela dinâmica do sistema. Na metodologia proposta, a lei de aprendizado inclui um termo de momentum que permite acelerar a aprendizagem dos valores das tabelas constantemente durante a execução dos testes, melhorando a resposta a cada evento. O desenvolvimento de um controle por aprendizado Neuro-Fuzzy foi motivado pela necessidade de ter um agente com a capacidade de aprendizado e armazenamento dos pontos ótimos de reversão. Este modelo de controle também consiste na implementação de um controle do tipo bang-bang, trabalhando com a servo-válvula sempre nos seus limites extremos de operação. O instante de reversão é determinado pelo sistema Neuro-Fuzzy, o qual tem como entradas a gama (dobro da amplitude) e o valor mínimo do carregamento solicitado. O processo de aprendizado é feito pelas atualizações dos pesos do sistema Neuro-Fuzzy, baseado nos erros obtidos durante a execução dos testes, melhorando a resposta do sistema a cada evento. A validação experimental dos modelos propostos é feita em uma máquina servohidráulica de ensaios de fadiga. Para este fim, o algoritmo de controle proposto foi implementado em tempo real em um módulo de controle CompactRIO da National Instruments. Os testes efetuados demonstraram a eficiência da metodologia proposta. / [en] In this thesis, accelerated learning and Neuro-Fuzzy control techniques were developed and applied to a servo-hydraulic system used in fatigue tests. This work involved four main stages: literature review, development of an accelerated learning control, development of a Neuro-Fuzzy control, and implementation of the learning control models into a fatigue testing machine. The accelerated learning control was implemented based on a learning control developed in previous works, introducing a faster learning law. Both learning control methodologies consist on implementing a bang-bang control, forcing the servovalve to always work in its operational limits. As the servo-valve works in its operational limits, the reversion points to achieve every peak or valley in the desired history are obtained by the learning algorithm, and stored in a specific table for each combination of minimum and mean load. The servo-valve reversion points depend on a few factors, such as alternate and mean loading components, while they are as well influenced by the system dynamics. In the proposed accelerated methodology, the learning law includes one momentum term that allows to speed up the learning process of the table cell values during the execution of the tests. The developed Neuro-Fuzzy control also consists on a bang-bang control, making the servo-valve work in its operational limits. However, here the instant of each reversion is determined by the Neuro-Fuzzy system, which has the load range and minimum load required as inputs. The learning process is made by the update of the Neuro-Fuzzy system weights, based on the errors obtained during the execution of the test.The experimental validation of the proposed models was made using a servo-hydraulic testing machine. The control algorithm was implemented in real time in a C-RIO computational system. The tests demonstrated the efficiency of the proposed methodology.

Made available in DSpace on 2014-12-17T14:54:55Z (GMT). No. of bitstreams: 1 MarconiCR_TESE.pdf: 2377871 bytes, checksum: c798a5eab76defef17ac0fe081e2453d (MD5) Previous issue date: 2010-03-16 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior / This paper presents a new multi-model technique of dentification in ANFIS for nonlinear systems. In this technique, the structure used is of the fuzzy Takagi-Sugeno of which the consequences are local linear models that represent the system of different points of operation and the precursors are membership functions whose adjustments are realized by the learning phase of the neuro-fuzzy ANFIS technique. The models that represent the system at different points of the operation can be found with linearization techniques like, for example, the Least Squares method that is robust against sounds and of simple application. The fuzzy system is responsible for informing the proportion of each model that should be utilized, using the membership functions. The membership functions can be adjusted by ANFIS with the use of neural network algorithms, like the back propagation error type, in such a way that the models found for each area are correctly interpolated and define an action of each model for possible entries into the system. In multi-models, the definition of action of models is known as metrics and, since this paper is based on ANFIS, it shall be denominated in ANFIS metrics. This way, ANFIS metrics is utilized to interpolate various models, composing a system to be identified. Differing from the traditional ANFIS, the created technique necessarily represents the system in various well defined regions by unaltered models whose pondered activation as per the membership functions. The selection of regions for the application of the Least Squares method is realized manually from the graphic analysis of the system behavior or from the physical characteristics of the plant. This selection serves as a base to initiate the linear model defining technique and generating the initial configuration of the membership functions. The experiments are conducted in a teaching tank, with multiple sections, designed and created to show the characteristics of the technique. The results from this tank illustrate the performance reached by the technique in task of identifying, utilizing configurations of ANFIS, comparing the developed technique with various models of simple metrics and comparing with the NNARX technique, also adapted to identification / Este trabalho apresenta uma nova t?cnica de identifica??o multimodelos baseada em ANFIS para sistemas n?o lineares. Nesta t?cnica, a estrutura utilizada ? do tipo fuzzy Takagi-Sugeno cujos consequentes s?o modelos lineares locais que representam o sistema em diferentes pontos de opera??o e os antecedentes s?o fun??es de pertin?ncia cujos ajustes s?o realizados pela fase de aprendizagem da t?cnica neuro-fuzzy ANFIS. Modelos que representem o sistema em diferentes pontos de opera??o podem ser encontrados com t?cnicas de lineariza??o como, por exemplo, o m?todo dos M?nimos Quadrados que ? robusto a ru?dos e de simples aplica??o. Cabe ? fase de implica??o do sistema fuzzy informar a propor??o de cada modelo que deve ser empregada, utilizando, para isto, as fun??es de pertin?ncia. As fun??es de pertin?ncia podem ser ajustadas pelo ANFIS com o uso de algoritmos de redes neurais, como o de retropropaga??o do erro, de modo que os modelos encontrados para cada regi?o sejam devidamente interpolados e, assim, definam-se a atua??o de cada modelo para as poss?veis entradas do sistema. Em multimodelos a defini??o de atua??o de modelos ? conhecida por m?trica e, como neste trabalho ? realizada pelo ANFIS, ser? denominada de m?trica ANFIS. Desta forma, uma m?trica ANFIS ? utilizada para interpolar v?rios modelos, compondo o sistema a ser identificado. Diferentemente do ANFIS tradicional, a t?cnica desenvolvida necessariamente representa o sistema em v?rias regi?es bem definidas por modelos inalter?veis que, por sua vez, ter?o sua ativa??o ponderada a partir das fun??es de pertin?ncia. A sele??o de regi?es para a aplica??o do m?todo dos M?nimos Quadrados ? realizada manualmente a partir da an?lise gr?fica do comportamento do sistema ou a partir do conhecimento de caracter?sticas f?sicas da planta. Esta sele??o serve como base para iniciar a t?cnica definindo modelos lineares e gerando a configura??o inicial das fun??es de pertin?ncia. Experimentos s?o realizados em um tanque did?tico, com m?ltiplas se??es, projetado e desenvolvido com a finalidade de mostrar caracter?sticas da t?cnica. Os resultados neste tanque ilustram o bom desempenho alcan?ado pela t?cnica na tarefa de identifica??o, utilizando, para isto, v?rias configura??es do ANFIS, comparando a t?cnica desenvolvida com m?ltiplos modelos de m?trica simples e comparando com a t?cnica NNARX, tamb?m adaptada para identifica??o

Práce popisuje základy principu funkčnosti neuronů a vytvoření umělých neuronových sítí. Je zde důkladně popsána struktura a funkce neuronů a ukázán nejpoužívanější algoritmus pro učení neuronů. Základy fuzzy logiky, včetně jejich výhod a nevýhod, jsou rovněž prezentovány. Detailněji je popsán algoritmus zpětného šíření chyb a adaptivní neuro-fuzzy inferenční systém. Tyto techniky poskytují efektivní způsoby učení neuronových sítí.

Práce popisuje základy principu funkčnosti neuronů a vytvoření umělých neuronových sítí. Je zde důkladně popsána struktura a funkce neuronů a ukázán nejpoužívanější algoritmus pro učení neuronů. Základy fuzzy logiky, včetně jejich výhod a nevýhod, jsou rovněž prezentovány. Detailněji je popsán algoritmus zpětného šíření chyb a adaptivní neuro-fuzzy inferenční systém. Tyto techniky poskytují efektivní způsoby učení neuronových sítí.

In the literature, various Mobile Ad hoc NETwork (MANET) routing protocols proposed. Each performs the best under specific context conditions, for example under high mobility or less volatile topologies. In existing MANET, the degradation in the routing protocol performance is always associated with changes in the network context. To date, no MANET routing protocol is able to produce optimal performance under all possible conditions. The core aim of this thesis is to solve the routing problem in mobile Ad hoc networks by introducing an optimum system that is in charge of the selection of the running routing protocol at all times, the system proposed in this thesis aims to address the degradation mentioned above. This optimisation system is a novel approach that can cope with the network performance’s degradation problem by switching to other routing protocol. The optimisation system proposed for MANET in this thesis adaptively selects the best routing protocol using an Artificial Intelligence mechanism according to the network context. In this thesis, MANET modelling helps in understanding the network performance through different contexts, as well as the models’ support to the optimisation system. Therefore, one of the main contributions of this thesis is the utilisation and comparison of various modelling techniques to create representative MANET performance models. Moreover, the proposed system uses an optimisation method to select the optimal communication routing protocol for the network context. Therefore, to build the proposed system, different optimisation techniques were utilised and compared to identify the best optimisation technique for the MANET intelligent system, which is also an important contribution of this thesis. The parameters selected to describe the network context were the network size and average mobility. The proposed system then functions by varying the routing mechanism with the time to keep the network performance at the best level. The selected protocol has been shown to produce a combination of: higher throughput, lower delay, fewer retransmission attempts, less data drop, and lower load, and was thus chosen on this basis. Validation test results indicate that the identified protocol can achieve both a better network performance quality than other routing protocols and a minimum cost function of 4.4%. The Ad hoc On Demand Distance Vector (AODV) protocol comes in second with a cost minimisation function of 27.5%, and the Optimised Link State Routing (OLSR) algorithm comes in third with a cost minimisation function of 29.8%. Finally, The Dynamic Source Routing (DSR) algorithm comes in last with a cost minimisation function of 38.3%.

Artificial intelligence techniques have been widely used for prediction in various areas of sciences and engineering. In the thesis, applications of AI techniques are studied to predict the dilution of industrial outfall discharges. The discharge of industrial effluents from the outfall systems is broadly divided into two categories on the basis of density. The effluent with density higher than the water receiving will sink and called as negatively buoyant jet. The effluent with density lower than the receiving water will rise and called as positively buoyant jet. The effluent discharge in the water body creates major environmental threats. In this work, negatively buoyant jet is considered. For the study, ANFIS model is taken into consideration and incorporated with algorithms such as GA, PSO and FFA to determine the suitable model for the discharge prediction. The training and test dataset for the ANFIS-type models are obtained by simulating the jet using the realizable k-ε turbulence model over a wide range of Froude numbers i.e. from 5 to 60 and discharge angles from 20 to 72.5 degrees employing OpenFOAM platform. Froude number and angles are taken as input parameters for the ANFIS-type models. The output parameters were peak salinity (Sm), return salinity (Sr), return point in x direction (xr) and peak salinity coordinates in x and y directions (xm and ym). Multivariate regression analysis has also been done to verify the linearity of the data using the same input and output parameters. To evaluate the performance of ANFIS, ANFIS-GA, ANFIS-PSO, ANFIS-FFA and multivariate regression model, some statistical parameters such as coefficient of determination (R2), root mean squared error (RMSE), mean absolute error (MAE) and average absolute deviation in percentage are determined. It has been observed that ANFIS-PSO is better in predicting the discharge characteristics.

Mitigation of displacement and acceleration responses of a three story benchmark structure excited by seismic motions is pursued in this study. Multiple 20-kN magnetorheological (MR) dampers are installed in the three-story benchmark structure and managed by a global fuzzy logic controller to provide smart damping forces to the benchmark structure. Two configurations of MR damper locations are considered to display multiple-input, single-output and multiple-input, multiple-output control capabilities. Characterization tests of each MR damper are performed in a laboratory to enable the formulation of fuzzy inference models. Prediction of MR damper forces by the fuzzy models shows sufficient agreement with experimental results. A controlled-elitist multi-objective genetic algorithm is utilized to optimize a set of fuzzy logic controllers with concurrent consideration to four structural response metrics. The genetic algorithm is able to identify optimal passive cases for MR damper operation, and then further improve their performance by intelligently modulating the command voltage for concurrent reductions of displacement and acceleration responses. An optimal controller is identified and validated through numerical simulation and fullscale experimentation. Numerical and experimental results show that performance of the controller algorithm is superior to optimal passive cases in 43% of investigated studies. Furthermore, the state-space model of the benchmark structure that is used in numerical simulations has been improved by a modified version of the same genetic algorithm used in development of fuzzy logic controllers. Experimental validation shows that the state-space model optimized by the genetic algorithm provides accurate prediction of response of the benchmark structure to base excitation.

The advances and the development of vehicles and autobalance robots make necessary the investigation of controllers able to meet the various challenges related to the use of these systems. The focus of this work is to study the equilibrium and position control of one two-wheeled robot. The particular interest in this application comes from its structure and its rich physical dynamics. Since this is a complex and non trivial problem, there is great interest in to analyze intelligent controllers. The first part of this dissertation discusses the development of a classic PID controller. Then it is compared with two types of intelligent controllers: On-line Neural Fuzzy Control (ONFC) and Proportional-Integral-Derivative Neural-Network (PID-NN). Also it is presented the implementation of controllers in a hadware plataform using the LEGO Mindstorm kit and in a simulation plataform using the MATLAB-Simulink. Two case studies are developed. The first one investigates the control of equilibrium and position of two-wheeled robot on a flat terrain to observe the intrinsec performance in lack of external factors. The second case studies the equilibrium and position control of the robot in irregular terrains to investigate the system response under influence of hard conditions in its environment. Finally, the performance of each controller developed is discussed and competitive results in the control of two-wheeled robot are achieved. / Com o avanço no desenvolvimento e utilização de veículos e robôs autoequilibrantes, faz-se necessário a investigação de controladores capazes de atender os diversos desafios relacionados à utilização desses sistemas. Neste trabalho foi estudado o controle de equilíbrio e posição de um robô auto-equilibrante de duas rodas. O interesse particular nesta aplicação vem da sua estrutura e da riqueza de sua dinâmica física. Por ser um problema complexo e não trivial há grande interesse em avaliar os controladores inteligentes. A primeira parte da dissertação aborda o desenvolvimento de um controle clássico do tipo PID, para em seguida ser comparado com a implementação de dois tipos de controladores inteligentes: On-line Neuro Fuzzy Control (ONFC) e Proportional-Integral-Derivative Neural-Network (PIDNN). Também é apresentada a implementação dos controladores em uma plataforma de hardware, utilizando o kit LEGO Mindstorm, e numa plataforma de simulação utilizando o MATLAB-Simulink. Em seguida, dois estudos de casos são desenvolvidos visando comparar o desempenho dos controladores. O primeiro caso avalia o controle de equilíbrio e posição do robô auto-equilibrante de duas rodas sobre um terreno plano tendo como interesse observar o desempenho intrínseco do sistema sob ausência de fatores externos. O segundo caso estuda o controle de equilíbrio e posição do robô em terrenos irregulares visando investigar a resposta do sistema sob influência de condições adversas em seu ambiente. Finalmente, o desempenho de cada um dos controladores desenvolvidos é discutido, verificando-se resultados competitivos no controle do robô auto-equilibrante de duas rodas.

Submitted by Maykon Nascimento (maykon.albani@hotmail.com) on 2016-06-27T19:26:14Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Dissertacao Giovanni Correia.pdf: 2352633 bytes, checksum: af81ecb43db7a1390cce952e53aaff53 (MD5) / Approved for entry into archive by Patricia Barros (patricia.barros@ufes.br) on 2016-06-28T12:18:13Z (GMT) No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Dissertacao Giovanni Correia.pdf: 2352633 bytes, checksum: af81ecb43db7a1390cce952e53aaff53 (MD5) / Made available in DSpace on 2016-06-28T12:18:13Z (GMT). No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Dissertacao Giovanni Correia.pdf: 2352633 bytes, checksum: af81ecb43db7a1390cce952e53aaff53 (MD5) / FAPES / Os modelos de árvores individuais são compostos por submodelos que estimam, geralmente, a competição, a mortalidade e o crescimento em diâmetro e altura de cada árvore. São usualmente adotados quando se deseja o melhor detalhamento da informação para estimar multiprodutos da floresta. Nesses modelos, as estimativas do crescimento em diâmetro a 1,30 m do solo (DAP) e a altura total (H) é obtida por meio de análise de regressão. Recentemente, técnicas de inteligência artificial estão sendo utilizadas com bom desempenho na mensuração florestal. Portanto, o objetivo desse trabalho foi avaliar o desempenho de técnicas de inteligência artificial (redes neurais artificiais e sistemas neuro-fuzzy) para estimar o crescimento em DAP e altura de árvores de eucalipto. Utilizou-se dados de inventários florestais contínuos de eucalipto, com medições anuais de DAP, altura total das 15 primeiras árvores da parcela e altura dominante, de acordo com o conceito de Assmann (1970), de 398 parcelas. O banco de dados foi dividido em 70% das parcelas para o treinamento das redes neurais artificiais e do sistema neuro-fuzzy; 15% das parcelas para a validação cruzada; e 15% das parcelas para validação dos sistemas. Com base nos resultados, notou-se que o índice de competição independente da distância 5 – IID5, proposto por Glover; Hool (1979), foi o que teve a maior correlação com as variáveis idade, crescimento em DAP e altura. Observou-se que as técnicas de inteligência artificial apresentaram boa precisão na estimativa do crescimento em DAP e altura total. As duas técnicas abordadas podem ser utilizadas para a prognose do DAP e altura total. / The models are composed of individual trees submodels estimating generally competition, mortality and growth height and diameter of each tree. Are usually adopted when you want the best detailed information to estimate forest multiproducts. In these models, estimates of growth in diameter at 1.30 m above the ground (DBH) and total height (H) is obtained by regression analysis. Recently, artificial intelligence techniques are being used with good performance in forest measurement. Therefore, the aim of this study was to evaluate the performance of artificial intelligence techniques (artificial neural networks and neuro-fuzzy systems) to estimate the growth in DAP and height of eucalyptus trees. We used continuous data eucalyptus forest inventories annually measurements DAP total height of the first 15 trees and dominant height of the portion, according to the concept of Assmann (1970), 398 parts. The database was divided into 70% of the plots for the training of artificial neural networks and neuro-fuzzy system; 15% of the plots for the cross-validation; and 15% of the plots for validating systems. Based on the results, it was noted that the independent competition index of distance 5 - IID5 proposed by Glover; Hool (1979), was the one that had the highest correlation with the age, growth in DAP and height. It was observed that the artificial intelligence techniques showed good accuracy in estimating the growth in DBH and total height. The two techniques discussed can be used for prognosis and overall height of DAP.