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.

Made available in DSpace on 2014-06-12T17:39:51Z (GMT). No. of bitstreams: 2 arquivo6971_1.pdf: 519832 bytes, checksum: 35de3846e1bc6e866dd2ee8b7a6bc74b (MD5) license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2006 / O presente trabalho apresenta um sistema de previsão de carga horária em curto prazo (sete dias à frente) formado por duas etapas. Na primeira etapa foram escolhidas duas redes neurais artificiais para prever o consumo diário total em um horizonte de sete dias à frente, uma rede para os dias úteis e outra para aos dias não-úteis, o processo de escolha das redes passou por uma análise da estrutura de entrada, da base de dados e do algoritmo de treinamento. Para gerar as melhores redes utilizou-se o método k-fold crossvalidation. A segunda etapa é responsável em fornecer o comportamento da curva de carga, ou seja, a distribuição horária do consumo diário, para isso utilizou-se o sistema ANFIS (Adaptive Network-based Fuzzy Inference System) para gerar um Sistema de Inferência Fuzzy- SIF que fornece um coeficiente que representa a fração do consumo horário em relação ao consumo diário, para inicialização dos modelos optou-se pela comparação entre dois métodos: o método de clusterização subtrativa desenvolvido por Chui S e o método por inspeção onde o SIF é gerado a partir do conhecimento do especialista. Optou-se por estes modelos devido à facilidade de implementação, a capacidade de generalização e resposta rápida. Os resultados obtidos foram comparados com a bibliografia e mostram que o modelo desenvolvido tem alta capacidade de generalização e apresenta baixos valores de MAPE (erro médio percentual), além de utilizar somente dados de carga elétrica como entrada para as redes e para o sistema ANFIS sem a necessidade de dados climáticos

Computerized adaptive testing attempts to provide the most suitable question for an examinee depending on the examinee¡¦s ability to achieve the best result. Although Maximum Likelihood Estimation (MLE) and Bayesian Likelihood Estimation (BLE) have been provided to solve ability estimation and have good results in the literature, little attention has been paid to the situation when the answer of an item does not conform with the examinee¡¦s ability as expected nor standard derivation changes of the ability estimation. We hypothesized that the Adaptive-Network-Based Fuzzy Inference System (ANFIS) can be used to infer flexible examinee¡¦s ability estimation automically by analyzing the relevant data of the examinee in a test. Consequently, the study presents a novel learning ability model based on ANFIS, which can adaptively choose questions by Item Response Theory. Taking the item discrimination, difficulty, guessing, and the examinee¡¦s ability before he/she answers a question as parameters, the proposed method can infer the adjustment of the examinee¡¦s ability to update its value after he/she answers the question. The ANFIS model of the experiments were developed using MATLAB. The examinees were simulated and the training data were collected under three different situations. Through different combination of ANFIS fuzzy rules, the adjustment of ability is inferred to improve the accuracy of the estimated ability. The error between the true ability and the estimated ability obtained by the proposed model is compared with MLE and BLE. The simulation results show that the estimated ability error of ANFIS is smaller than MLE and BLE when the value of the test information is larger. The proposed method could provide better accuracy of the examinee¡¦s ability and offer more appropriate questions for examinees. Keywords: ANFIS, Item Response Theory, Computerized Adaptive Testing

Este trabalho apresenta o desenvolvimento de um sistema inteligente para análise da técnica de pedalada aplicada por ciclistas. Para isso, desenvolveu-se um par de pedais de encaixe instrumentados, a partir dos quais é possível medir a componente de força normal aplicada nas partes frontal e posterior dos pedais. O modelo virtual da célula de carga experimental foi desenvolvido através da digitalização dos pedais de encaixe comerciais, utilizando-se um sistema comercial de escaneamento 3D com precisão declarada de 0,1mm. Cada pedal foi instrumentado com oito extensômetros de resistência elétrica (HBM 1-LY-13-1.5/350). Posteriormente os carregamentos máximos em cada eixo de medida de força foram estabelecidos utilizando-se uma plataforma de aquisição comercial específica para medida de deformação mecânica. Considerando-se os valores determinados, desenvolveu-se o circuito de condicionamento e realizaram-se os ensaios de deformação estática, obtendo-se as funções de transferência de saída de tensão elétrica em função do carregamento mecânico. O erro de linearidade máximo, considerando todos os canais, ficou abaixo de 0,75% e a máxima incerteza expandida (k=2) por canal, obtida através da aplicação do método clássico, foi de 1,55%. Em sequência, integrouse o sistema de pedais desenvolvido a dois outros sistemas, são eles: um par pedivelas experimentais instrumentados, capazes de medir as três componentes da força aplicada aos pedais e transmitidas aos pedivelas com um erro de linearidade abaixo de 0,6% e uma incerteza combinada inferior a 3,22%, e um sistema de cinemetria comercial, cuja precisão declarada pelo fabricante é de 1mm. Para possibilitar uma comparação quantitativa entre treinos ou ciclistas, implementou-se um sistema inteligente, baseado em redes Neuro-Fuzzy (ANFIS). A partir dos valores da potência média, do desvio padrão da potência e da assimetria bilateral média, obtidos ao longo de ensaios realizados sob protocolo desenvolvido especificamente para este trabalho, um score que representa o nível da técnica de pedalada apresentado pelo ciclista é determinado. Com intuito de testar o sistema, desenvolveu-se um projeto de experimentos com 2 fatores controláveis (sujeito e nível de frenagem de um rolo de treinamento), e realizou-se ensaios com oito ciclistas de características fisiológicas e níveis de preparos distintos. Através da análise estatística, constatou-se que das 23 variáveis de resposta consideradas ao longo do experimento, 23 são influenciadas significativamente pelo fator controlado sujeito e oito são influenciadas significativamente pelo fator controlado nível de frenagem magnética. / This report describes the development of an intelligent pedaling technique analysis system. To accomplish that, a pair of road bicycle pedals (SHIMANO R540) were instrumented to measure the forces that are applied to the front and back regions of the pedals. The virtual models of the pedals were developed based on a 3D scanned mesh developed with aid of a commercial 3D scanning system with a precision of 0.1mm. Each pedal was instrumented with eight electrical resistance strain-gages (HBM 1-LY-13-1.5/350). After that, the range of the mechanical deformation of each measurement channel was determined with aid of an industrial deformation acquisition system. The conditioning circuit was developed based on the mechanical deformation ranges previously determined and the static calibration experiment was performed to determine the voltage output transfer functions. The maximum linearity error determined per channel was 0,75% and the maximum expanded uncertainty (k=2), determined applying the classical methodology, was 1,55%. After that, the instrumented pedals developed were integrated with two complementary systems, which are: a pair of instrumented crank arm load cells which measure the components of the force applied to the bicycle pedal with a linearity error under 0.6% and an uncertainty of 3,22% and an Optitrack motion track system with a declared accuracy of 1mm. An intelligent pedaling technique analysis system was implemented through an Adaptive Neuro Fuzzy Inference System (ANFIS) to determine the cyclist pedaling technique score based on three inputs: the average power applied to bicycle pedal, the average power standard deviation and the bilateral asymmetry index, all of them collected under an experimental protocol specifically designed for this application. To evaluate the behavior of the system developed a randomized block experiment design with two controlled factors was performed indoor with aid of an ergometer roll; 160 sprints were conducted with eight subjects of different training levels. From the data collected an ANOVA test was performed, which confirmed that all the 23 response variables vary significantly in function of the subject’s controlled factor and eight of them vary significantly in function of the magnetic braking level.

Nowadays, industrial processes are having to be rapidly developed to meet high standards regarding increases in the production rate and/or improving product quality. Fulfilling these requirements is having to work in tandem with the pressure to reduce energy consumption due to global environmental regulations. Consequently, most industrial processes critically rely on automatic control, which can provide efficient solutions to meet such challenges and prerequisites. For this thesis, an intelligent system design has been investigated for controlling the distillation process, which is characterised by highly nonlinear and dynamic behaviour. These features raise very challenging tasks for control systems designers. Fuzzy logic and artificial neural networks (ANNs) are the main methods used in this study to design different controllers, namely: PI- PD- and PID-like fuzzy controllers, ANN-based NARMAL2 in addition to a conventional PID controller for comparison purposes. Genetic algorithm (GA) and particle swarm optimisation (PSO) have also been utilised to tune fuzzy controllers by finding the best set of scaling factors. Finally, an intelligent controller is proposed, called ANFIS-based NARMA-L2, which uses ANFIS as an approximation approach for identifying the underlying systems in a NARMA-L2 configuration. The controllers are applied to control two compositions of a binary distillation column, which has been modelled and simulated in MATLAB® and on the Simulink® platform. Comparative analysis has been undertaken to investigate the controllers' performance, which shows that PID-like FLC outperforms the other tested fuzzy control configurations, i.e. PI- and PD-like. Moreover, PSO has been found to outperform GA in finding the best set of scaling factors and over a shorter time period. Subsequently, the performance of PID-like FLC has been compared with ANN-based NARMA-L2 and the proposed ANFIS-based NARMA-L2, by subjecting the controlled column to different test scenarios. Furthermore, the stability and robustness of the controllers have been assessed by subjecting the controlled column to inputs variance and disturbances situations. The proposed ANFIS-based NARMAL2 controller outperforms and demonstrates more tolerance of disturbances than the other controllers. Finally, the study has involved investigating the control of a multicomponent distillation column due to its significant enhancement in operational efficiency regarding energy saving and recent widespread implementation. That is, Kaibel's distillation column with 4×4 configuration has been simulated also in MATLAB® and on the Simulink® platform with the proposed controller being implemented to control the temperatures of the column and the outcomes subsequently compared with conventional PID controllers. Again, the novel controller has proven its superiority regarding the disturbances tolerance as well as dealing with the high dynamics and nonlinear behaviour.

Este trabalho apresenta o desenvolvimento de um sistema inteligente para análise da técnica de pedalada aplicada por ciclistas. Para isso, desenvolveu-se um par de pedais de encaixe instrumentados, a partir dos quais é possível medir a componente de força normal aplicada nas partes frontal e posterior dos pedais. O modelo virtual da célula de carga experimental foi desenvolvido através da digitalização dos pedais de encaixe comerciais, utilizando-se um sistema comercial de escaneamento 3D com precisão declarada de 0,1mm. Cada pedal foi instrumentado com oito extensômetros de resistência elétrica (HBM 1-LY-13-1.5/350). Posteriormente os carregamentos máximos em cada eixo de medida de força foram estabelecidos utilizando-se uma plataforma de aquisição comercial específica para medida de deformação mecânica. Considerando-se os valores determinados, desenvolveu-se o circuito de condicionamento e realizaram-se os ensaios de deformação estática, obtendo-se as funções de transferência de saída de tensão elétrica em função do carregamento mecânico. O erro de linearidade máximo, considerando todos os canais, ficou abaixo de 0,75% e a máxima incerteza expandida (k=2) por canal, obtida através da aplicação do método clássico, foi de 1,55%. Em sequência, integrouse o sistema de pedais desenvolvido a dois outros sistemas, são eles: um par pedivelas experimentais instrumentados, capazes de medir as três componentes da força aplicada aos pedais e transmitidas aos pedivelas com um erro de linearidade abaixo de 0,6% e uma incerteza combinada inferior a 3,22%, e um sistema de cinemetria comercial, cuja precisão declarada pelo fabricante é de 1mm. Para possibilitar uma comparação quantitativa entre treinos ou ciclistas, implementou-se um sistema inteligente, baseado em redes Neuro-Fuzzy (ANFIS). A partir dos valores da potência média, do desvio padrão da potência e da assimetria bilateral média, obtidos ao longo de ensaios realizados sob protocolo desenvolvido especificamente para este trabalho, um score que representa o nível da técnica de pedalada apresentado pelo ciclista é determinado. Com intuito de testar o sistema, desenvolveu-se um projeto de experimentos com 2 fatores controláveis (sujeito e nível de frenagem de um rolo de treinamento), e realizou-se ensaios com oito ciclistas de características fisiológicas e níveis de preparos distintos. Através da análise estatística, constatou-se que das 23 variáveis de resposta consideradas ao longo do experimento, 23 são influenciadas significativamente pelo fator controlado sujeito e oito são influenciadas significativamente pelo fator controlado nível de frenagem magnética. / This report describes the development of an intelligent pedaling technique analysis system. To accomplish that, a pair of road bicycle pedals (SHIMANO R540) were instrumented to measure the forces that are applied to the front and back regions of the pedals. The virtual models of the pedals were developed based on a 3D scanned mesh developed with aid of a commercial 3D scanning system with a precision of 0.1mm. Each pedal was instrumented with eight electrical resistance strain-gages (HBM 1-LY-13-1.5/350). After that, the range of the mechanical deformation of each measurement channel was determined with aid of an industrial deformation acquisition system. The conditioning circuit was developed based on the mechanical deformation ranges previously determined and the static calibration experiment was performed to determine the voltage output transfer functions. The maximum linearity error determined per channel was 0,75% and the maximum expanded uncertainty (k=2), determined applying the classical methodology, was 1,55%. After that, the instrumented pedals developed were integrated with two complementary systems, which are: a pair of instrumented crank arm load cells which measure the components of the force applied to the bicycle pedal with a linearity error under 0.6% and an uncertainty of 3,22% and an Optitrack motion track system with a declared accuracy of 1mm. An intelligent pedaling technique analysis system was implemented through an Adaptive Neuro Fuzzy Inference System (ANFIS) to determine the cyclist pedaling technique score based on three inputs: the average power applied to bicycle pedal, the average power standard deviation and the bilateral asymmetry index, all of them collected under an experimental protocol specifically designed for this application. To evaluate the behavior of the system developed a randomized block experiment design with two controlled factors was performed indoor with aid of an ergometer roll; 160 sprints were conducted with eight subjects of different training levels. From the data collected an ANOVA test was performed, which confirmed that all the 23 response variables vary significantly in function of the subject’s controlled factor and eight of them vary significantly in function of the magnetic braking level.

Este trabalho apresenta o desenvolvimento de um sistema inteligente para análise da técnica de pedalada aplicada por ciclistas. Para isso, desenvolveu-se um par de pedais de encaixe instrumentados, a partir dos quais é possível medir a componente de força normal aplicada nas partes frontal e posterior dos pedais. O modelo virtual da célula de carga experimental foi desenvolvido através da digitalização dos pedais de encaixe comerciais, utilizando-se um sistema comercial de escaneamento 3D com precisão declarada de 0,1mm. Cada pedal foi instrumentado com oito extensômetros de resistência elétrica (HBM 1-LY-13-1.5/350). Posteriormente os carregamentos máximos em cada eixo de medida de força foram estabelecidos utilizando-se uma plataforma de aquisição comercial específica para medida de deformação mecânica. Considerando-se os valores determinados, desenvolveu-se o circuito de condicionamento e realizaram-se os ensaios de deformação estática, obtendo-se as funções de transferência de saída de tensão elétrica em função do carregamento mecânico. O erro de linearidade máximo, considerando todos os canais, ficou abaixo de 0,75% e a máxima incerteza expandida (k=2) por canal, obtida através da aplicação do método clássico, foi de 1,55%. Em sequência, integrouse o sistema de pedais desenvolvido a dois outros sistemas, são eles: um par pedivelas experimentais instrumentados, capazes de medir as três componentes da força aplicada aos pedais e transmitidas aos pedivelas com um erro de linearidade abaixo de 0,6% e uma incerteza combinada inferior a 3,22%, e um sistema de cinemetria comercial, cuja precisão declarada pelo fabricante é de 1mm. Para possibilitar uma comparação quantitativa entre treinos ou ciclistas, implementou-se um sistema inteligente, baseado em redes Neuro-Fuzzy (ANFIS). A partir dos valores da potência média, do desvio padrão da potência e da assimetria bilateral média, obtidos ao longo de ensaios realizados sob protocolo desenvolvido especificamente para este trabalho, um score que representa o nível da técnica de pedalada apresentado pelo ciclista é determinado. Com intuito de testar o sistema, desenvolveu-se um projeto de experimentos com 2 fatores controláveis (sujeito e nível de frenagem de um rolo de treinamento), e realizou-se ensaios com oito ciclistas de características fisiológicas e níveis de preparos distintos. Através da análise estatística, constatou-se que das 23 variáveis de resposta consideradas ao longo do experimento, 23 são influenciadas significativamente pelo fator controlado sujeito e oito são influenciadas significativamente pelo fator controlado nível de frenagem magnética. / This report describes the development of an intelligent pedaling technique analysis system. To accomplish that, a pair of road bicycle pedals (SHIMANO R540) were instrumented to measure the forces that are applied to the front and back regions of the pedals. The virtual models of the pedals were developed based on a 3D scanned mesh developed with aid of a commercial 3D scanning system with a precision of 0.1mm. Each pedal was instrumented with eight electrical resistance strain-gages (HBM 1-LY-13-1.5/350). After that, the range of the mechanical deformation of each measurement channel was determined with aid of an industrial deformation acquisition system. The conditioning circuit was developed based on the mechanical deformation ranges previously determined and the static calibration experiment was performed to determine the voltage output transfer functions. The maximum linearity error determined per channel was 0,75% and the maximum expanded uncertainty (k=2), determined applying the classical methodology, was 1,55%. After that, the instrumented pedals developed were integrated with two complementary systems, which are: a pair of instrumented crank arm load cells which measure the components of the force applied to the bicycle pedal with a linearity error under 0.6% and an uncertainty of 3,22% and an Optitrack motion track system with a declared accuracy of 1mm. An intelligent pedaling technique analysis system was implemented through an Adaptive Neuro Fuzzy Inference System (ANFIS) to determine the cyclist pedaling technique score based on three inputs: the average power applied to bicycle pedal, the average power standard deviation and the bilateral asymmetry index, all of them collected under an experimental protocol specifically designed for this application. To evaluate the behavior of the system developed a randomized block experiment design with two controlled factors was performed indoor with aid of an ergometer roll; 160 sprints were conducted with eight subjects of different training levels. From the data collected an ANOVA test was performed, which confirmed that all the 23 response variables vary significantly in function of the subject’s controlled factor and eight of them vary significantly in function of the magnetic braking level.

Estimation of evaporation from open water is essential for hydrodynamics, manufacturing industries, irrigation, farming, environmental protection and many other purposes. It is also important for proper management of hydrological resources such as reservoirs, lakes and rivers. Recent methods are mostly data-driven methods, such as using Artificial Intelligence techniques. Adaptive Neuro Fuzzy Inference System (ANFIS) is one of them and has been widely adopted in many hydrological fields for its simplicity. The current research presents a comparative study on the impact of optimization techniques such as Firefly Algorithm (FFA), Genetic Algorithm (GA), Particle Swarm Optimizer (PSO) and Ant Colony Optimization (ACO) on obtained results. In addition, a practical method named Multi Gene-genetic Programming (MGGP) is employed to propose an equation for the estimation of the Evaporation. Six different measured weather variables are taken, which are maximum, minimum and average air temperature, sunshine hours, wind speed and relative humidity. Models are separately calibrated with total data set collected over an eight-year period of 2010-2017 at the specified station “Arizona” in the United States of America. Ten statistical indices are calculated to verify the results. All optimizers were observed and compared to check if the results are better than ANFIS or not. The objectives of the adoption of different optimizer techniques was to verify the accuracy of the prediction by ANFIS model. Comparisons showed that ANFIS and MGGP are slightly better than the other models. MGGP model is different from other models in a way that it provides a set of equations instead of showing numerical values; therefore, the computational time is high. PSO, FFA, ACO and GA are considered as optimizers in the main model. Though PSO provided very similar results to the ANFIS model and MGGP gives even better results than basic ANFIS model. ANFIS is easier in terms of model formation. ANFIS is simpler to build and easy to operate. Since the prediction was quite identical in all cases, the ANFIS model was suggested due to its simplicity.

Yes / Corneal images can be acquired using confocal microscopes which provide detailed images of the different layers inside the cornea. Most corneal problems and diseases occur in one or more of the main corneal layers: the epithelium, stroma and endothelium. Consequently, for automatically extracting clinical information associated with corneal diseases, or evaluating the normal cornea, it is important also to be able to automatically recognise these layers easily. Artificial intelligence (AI) approaches can provide improved accuracy over the conventional processing techniques and save a useful amount of time over the manual analysis time required by clinical experts. Artificial neural networks (ANN) and adaptive neuro fuzzy inference systems (ANFIS), are powerful AI techniques, which have the capability to accurately classify the main layers of the cornea. The use of an ANFIS approach to analyse corneal layers is described for the first time in this paper, and statistical features have been also employed in the identification of the corneal abnormality. An ANN approach is then added to form a combined committee machine with improved performance which achieves an accuracy of 100% for some classes in the processed data sets. Three normal data sets of whole corneas, comprising a total of 356 images, and seven abnormal corneal images associated with diseases have been investigated in the proposed system. The resulting system is able to pre-process (quality enhancement, noise removal), classify (whole data sets, not just samples of the images as mentioned in the previous studies), and identify abnormalities in the analysed data sets. The system output is visually mapped and the main corneal layers are displayed. 3D volume visualisation for the processed corneal images as well as for each individual corneal cell is also achieved through this system. Corneal clinicians have verified and approved the clinical usefulness of the developed system especially in terms of underpinning the expertise of ophthalmologists and its applicability in patient care.

Yes / This paper presents an investigation into the development of an intelligent active vibration control (AVC) system. Evolutionary Genetic algorithms (GAs) and Adaptive Neuro-Fuzzy Inference system (ANFIS) algorithms are used to develop mechanisms of an AVC system, where the controller is designed on the basis of optimal vibration suppression using the plant model. A simulation platform of a flexible beam system in transverse vibration using finite difference (FD) method is considered to demonstrate the capabilities of the AVC system using GAs and ANFIS. MATLAB GA tool box for GAs and Fuzzy Logic tool box for ANFIS function are used for AVC system design. The system is then implemented, tested and its performance assessed for GAs and ANFIS based design. Finally a comparative performance of the algorithm in implementing AVC system using GAs and ANFIS is presented and discussed through a set of experiments.