[pt] Este trabalho apresenta procedimentos de modelagem e controle neuro-fuzzy de sistemas dinâmicos. Neste contexto, é proposta e avaliada a utilização simultânea da abordagem neuro-fuzzy em todo o sistema de malha fechada controlador-planta.Na modelagem da planta, o espaço de entrada do sistema dinâmico é inicialmente dividido em um número de regiões de operação fuzzy onde modelos de ordem reduzida (ARMAX) representam o comportamento do sistema dinâmico. A saída completa do sistema - modelo global - é obtida através da conjunção das saídas dos modelos locais usando uma rede neuro-fuzzy.No controle da planta, é proposto um novo controlador neuro-fuzzy chamado Controlador Neuro-fuzzy de Coeficientes Variáveis (CNFCV), que tem como objetivos melhorar a robustez do sistema de controle a perturbações e a geração automática da variável manipulada, que é uma dificuldade normalmente encontrada em controladores neurais ou neuro-fuzzy. Esse controlador é originado dos modelos de redes neurais de Mellem (1997) e Velloso (1999), e utiliza redes neuro-fuzzy para a geração dos coeficientes variáveis de um modelo ARMA da variável manipulada. Apesar de juntar modelos de séries temporais com a abordagem neuro-fuzzy, o CNFCV tem como função não a previsão, mas sim o controle de uma planta ou processo.Para avaliar o desempenho do CNFCV são utilizados, como meios de comparação,controladores neuro- fuzzy conhecidos - FALCON-H Fuzzy Adaptive Learning Control Network with Hybrid Learning e NEFCON Neuro-Fuzzy Controller - e o tradicional controlador PID Proporcional- Integral-Derivativo.As plantas utilizadas são uma planta linear Bobinador, uma planta linearizada Pêndulo Invertido e uma planta não linear %CO2. A escolha de tais plantas deve-se ao fato de serem utilizadas e modeladas em aplicações práticas e em trabalhos acadêmicos. Os resultados obtidos com o CNFCV são analisados e comparados aos proporcionados pelas outras estruturas.Ao final são apresentadas conclusões e sugestões para trabalhos futuros. / [en] In this work procedures for neuro-fuzzy modelling and control of dynamic systems are reviewed and a new structure is proposed. In this, modelling and closed-loop control are performed simultaneously by using a neuro-fuzzy approach. In the modelling stage the input space of a dynamic system (plant) is initially divided into a number of fuzzy operating regions within which reduced order models are able to represent the system. The complete system model output - the global model - is obtained through the conjunction of the outputs of the local models. A new structure, called Neuro-Fuzzy Controller with Variable Coefficients (NFCVC) is proposed and evaluated. Its main objectives are to improve the system s robustness and to provide automatic generation of the manipulated variable in order to overcome a difficulty of neural and neuro-fuzzy controllers in general. The NFCVC is originated from models proposed by Mellem (1997) and Velloso (1999) and makes use of neuro-fuzzy networks to generate variable coefficients of an ARMA model. Despite combining times series models with a neuro-fuzzy approach, the main function of NFCVC is to perform the control of the plant.In order to evaluate the performance of NFCVC two well-known neuro-fuzzy controllers - FALCON-H (Fuzzy Adaptive Learning Control Network with Hybrid Learning) and the NEFCON (Neuro-Fuzzy Controller) - as well as the traditional PID controller are used as means of comparison.A linear plant (Rotor Winder), a linearized plant (Inverted Pendulum) and a nonlinear plant (%CO2) are used in the experiments. These plants are well-known and generally used in practical applications and/or academic works. The results for the NFCVC are analyzed and compared to those obtained with the others structures. Finally, conclusions and suggestions for future work are presented.

[pt] O presente trabalho é proposto o atributo PAC-MFCC operando com Classificadores em Sub-Bandas para a tarefa de identificação de locutor independente do texto em ruído. O sistema proposto é comparado com os atributos MFCC (Coeficientes Cepestrais de Frequência Mel), PAC- MFCC (Fase Autocorrelação-MFCC ) sem uso de classificadores em sub-bandas, SSCH(Histogramas de Centróides de Sub-Bandas Espectrais) e TECC (Coeficientes Cepestrais da Energia Teager). Nesta tarefa de reconhecimento, utilizou-se a base TIMIT a qual é composta de 630 locutores onde cada um deles falam 10 frases de aproximadamente 3 segundos cada frase, das quais 8 frases foram utilizadas para treinamento e 2 para teste, obtendo-se um total de 1260 locuções para o reconhecimento. Investigou-se o desempenho dos diversos sistemas utilizando diferentes tipos de ruídos da base Noisex 92 com diferentes relação sinal ruído. Verificou-se que a taxa de acerto da técnica PAC-MFCC com classificador em Sub-Bandas apresenta o melhor desempenho em comparação com as outras técnicas quando se tem uma relação sinal ruído menor que 10dB. / [en] In this work is proposed the use of the PAC-MFCC feature with Sub-band Classifiers for the task of text-independent speaker identification in noise. The proposed scheme is compared with the features MFCC (Mel-Frequency Cepstral Coefficients ), PAC-MFCC (Phase Autocorrelation MFCC) without subband classifiers, SSCH (Subband Spectral Centroid Histograms) and TECC (Teager Energy Cepstrum Coefficients). In this recognition task, we used the TIMIT database which consists of 630 speakers, where every one of them speak 10 utterances of 3 seconds each one approximately, of which eight utterance were used for training and two for testing, thus obtaining a total of 1260 test utterance for the recognition. We investigated the performance of these techniques using differents types of noise from the base Noisex 92 with different signal to noise ratios. It was found that the accuracy rate of the PAC-MFCC feature with Sub-band Classifiers performs better in comparison with other techniques at a lower signal noise(less than 10dB).

Concevoir un bâtiment avec des préoccupations de confort et de maitrise de l'énergie est un enjeu primordial. Dans le contexte mondial alarmant, tant sur le plan énergétique que climatique, le secteur du bâtiment est pointé du doigt. En France, l'éclairage fait partie des mauvais élèves comparés aux autres pays européens. Le pays accuse un certain retard dans la qualité de l'éclairage tertiaire ou résidentiel en termes de performance énergétique, confort visuel ou santé au travail. L'organisation de « 2015, Année de la Lumière en France » par l'ONU, l'UNESCO et le CNOP montre qu'un changement tend à s'opérer dans les consciences. L'île de La Réunion, compte tenu de son fort gisement solaire, est une terre de choix pour expérimenter des dispositifs novateurs dans l'apport de lumière naturelle au sein des bâtiments. Pour cela, l'approche physique du phénomène offre des perspectives de prédiction où l'étude théorique d'aujourd’hui fera le bâtiment de demain. En débutant par une approche généraliste, ce manuscrit de thèse se veut être le plus exhaustif possible dans l'étude bibliographique, expérimentale et modélisation du comportement des DGLT-R (Dispositifs de Guides Lumineux Tubulaires Réfléchissants) appliqués au bâtiment. Notre prospection bibliographique nous a révélé que ce procédé est couramment commercialisé et installé à l'échelle internationale sur les bâtiments soucieux de leur impact environnemental ou du bien-être de leurs occupants. Cependant, la littérature montre également qu'ils sont peu étudiés et que les approches ne peuvent être appliquées à tous types de climats. Dans ce cadre, une étude expérimentale à échelle réelle, comprenant de nombreux scénarios, a été menée en vue de créer une base de données de mesures expérimentales unique et comprendre les phénomènes mis en jeu. La modélisation semi-empirique possède à la fois des atouts et des inconvénients. Forte de simplicité, elle reste toutefois limitée à une configuration ou un climat particulier. À cette problématique, ces travaux de recherche offrent une solution : un outil numérique de modélisation dédié, assurant la génération des modèles au départ de mesures à effectuer, nommé HEMERA. Il utilise la puissance des algorithmes génétiques pour offrir une solution efficace à un problème local. Le code de calcul CODYRUN est le support de validation du modèle généré pour le climat de La Réunion. Ce code de calcul bénéficie d'une expérience et d'un développement de plus de 20 ans au sein du laboratoire PIMENT. / Designing a building with comfort and energy conservation concerns is a key issue. In the alarming global context, both on energy-climate that the building sector is singled out. In France, the lighting is among the poor performers compared to other European countries. The country lags behind in quality tertiary or residential lighting in terms of energy performance, visual comfort or health. The organization of "2015, Année de la Lumière en France" by the ONU, UNESCO and the CNOP shows a change which tends to take place in the popular consciousness. Reunion Island, considering its strong solar radiation, is a land of choice to experiment with innovative devices in the contribution of daylight in buildings. For this purpose, the physical approach of the phenomenon provides prediction prospects, where the theoretical study of today will be tomorrow's building. Starting with a generalist approach, this PhD thesis aims to be as comprehensive as possible in the literature review, experimental and modelling study of the TDGS (Tubular Daylight Guide Systems) applied to the building. Our bibliographic survey has revealed that, this process is commonly marketed and installed internationally on the buildings concerned about their environmental impact or well-being of their occupants. However, the literature also shows they are poorly studied, and the approaches cannot be applied to all types of climates. In this context, a full-scale experimental study, including many scenarios was conducted to understand the phenomena involved.The semi-empirical modelling has both strengths and weaknesses. With simplicity, it remains limited to a configuration or a particular climate. In this issue, this research offers a solution: a numeric tool of dedicated modelling environment, ensuring the generation of the models according to need, called HEMERA. It uses the power of genetic algorithms to provide an effective solution to a local problem. The computer code CODYRUN is the support of the validation of the model developed as part of this work, and adjusted specifically for the climate of Reunion Island. This software profits an experience and a development of over 20 years in the PIMENT laboratory.

No description available.

La pomme (Malus Domestica Borkh.), fruit largement répandu dans les pays tempérés est beaucoup consommée. Elle représente une source importante en composés phénoliques. Cette étude s’est intéressée aux polyphénols des tissus du parenchyme. La problématique s’oriente sur les effets de la texture sur la diffusion de ces molécules. L’originalité de l’approche repose sur l’association de la texture, de la pression osmotique et la diffusion des polyphénols. Les méthodes de caractérisations physiques et biochimiques ont permis de mesurer les changements à l´échelle macroscopique et les modifications chimiques qui s’opèrent dans les matrices végétales. Les résultats de l’étude du transfert de matière ont permis de mettre en évidence les différents facteurs pouvant influer sur les valeurs des coefficients de diffusion. La texture, l’épaisseur, la variété du fruit et la pression du milieu diffusant, constituent des facteurs pouvant influencer le transfert de matière. L’étude de l’évolution de composant de la paroi a montré des changements qui s’opèrent au cours de la diffusion. Des analyses microscopiques ont relevé les modifications à l’échelle cellulaire de la diffusion de procyanidines, polyphénols majoritaires et des interactions avec les composants pariétaux. / Apple (Malus domestica Borkh. ) fruit widespread in temperate countries, is much consumed.It represents an important source of phenolic compounds. This study was interestedin polyphenol content of apple tissue parenchyma. The problem concerns effects of texturedegradation on the diffusion of polyphenols molecules. The originality of the approach isbased on the combination of texture, osmotic pressure and polyphenol leaching. Physicaland biochemical methods were used to measure changes at macroscopic scale and chemicalchanges occurring in the parenchymateous tissue . The study of mass transfer highlightedvarious factors that may affect apparent coefficient diffusion. The result showed that thedisintegration of texture , thickness, apple variety and osmotic pressure of leaching mediacan influence mass transfer yield. The study of the Cell walls components showed changesthat occur during leaching process. Light microscopic analysis revealed changes at cellularscale, procyanidins the major polyphenols, leaching phenomena and also interactionswith cell walls matrix.

The submitted diploma thesis deals with the tax authority of municipalities in the area of the real estate tax. The aim of the thesis is to find out the measure of using real estate tax coefficients by the individual size of municipalities and then analyse reasons for their using or wasting. The analysis of using real estate tax coefficients and reasons of their using or wasting was performed in two regions, Jihočeský and Ústecký. The data was collected by questionnaires on the chosen sample of municipalities in the individual size of municipalities. The secondary data was collected from statistic published by the General financial headquarters. The gained data shows that some of real estate tax coefficients are used by 52 % of municipalities. However, more municipalities using the possibility of putting coefficients in place are located in Ústecký region. The least coefficients are used by municipalities in the size of ones up to 2 000 inhabitants, which are from all the groups the most influenced by the opinions of inhabitants.

The master thesis deals with national accounts, particularly with symmetric input-output tables and utilisation for research of investments change impact. The main aim is to show national accounts exploitation for economical modelling and a manual for modelling of concrete impact. This thesis concentrates on the topic of calculation of static model with application on real data of Czech Republic. The base of calculation is technical coefficients analyse.

Les modèles physico-chimiques destinés à représenter la réalité expérimentale peuvent se révéler inadéquats. C'est le cas du piège à oxyde d'azote, utilisé comme support applicatif de notre thèse, qui est un système catalytique traitant les émissions polluantes du moteur Diesel. Les sorties sont des courbes de concentrations des polluants, qui sont des données fonctionnelles, dépendant de concentrations initiales scalaires.L'objectif initial de cette thèse est de proposer des plans d'expériences ayant un sens pour l'utilisateur. Cependant les plans d'expérience s'appuyant sur des modèles, l'essentiel du travail a conduit à proposer une représentation statistique tenant compte des connaissances des experts, et qui permette de construire ce plan.Trois axes de recherches ont été explorés. Nous avons d'abord considéré une modélisation non fonctionnelle avec le recours à la théorie du krigeage. Puis, nous avons pris en compte la dimension fonctionnelle des réponses, avec l'application et l'extension des modèles à coefficients variables. Enfin en repartant du modèle initial, nous avons fait dépendre les paramètres cinétiques des entrées (scalaires) à l'aide d'une représentation non paramétrique.Afin de comparer les méthodes, il a été nécessaire de mener une campagne expérimentale, et nous proposons une démarche de plan exploratoire, basée sur l’entropie maximale. / Physico-chemical models designed to represent experimental reality may prove to be inadequate. This is the case of nitrogen oxide trap, used as an application support of our thesis, which is a catalyst system treating the emissions of the diesel engine. The outputs are the curves of concentrations of pollutants, which are functional data, depending on scalar initial concentrations.The initial objective of this thesis is to propose experiental design that are meaningful to the user. However, the experimental design relying on models, most of the work has led us to propose a statistical representation taking into account the expert knowledge, and allows to build this plan.Three lines of research were explored. We first considered a non-functional modeling with the use of kriging theory. Then, we took into account the functional dimension of the responses, with the application and extension of varying coefficent models. Finally, starting again from the original model, we developped a model depending on the kinetic parameters of the inputs (scalar) using a nonparametric representation.To compare the methods, it was necessary to conduct an experimental campaign, and we propose an exploratory design approach, based on maximum entropy.

O desenvolvimento da tecnologia de microarray tornou possível a mediçao dos níveis de expressão de centenas ou até mesmo milhares de genes simultaneamente para diversas condições experimentais. A grande quantidade de dados disponível gerou a demanda por métodos computacionais que permitam sua análise de forma eficiente e automatizada. Em muitos dos métodos computacionais empregados durante a análise de dados de expressão gênica é necessária a escolha de uma medida de proximidade apropriada entre genes ou amostras. Dentre as medidas de proximidade disponíveis, coeficientes de correlação têm sido amplamente empregados, em virtude da sua capacidade em capturar similaridades entre tendências das sequências numéricas comparadas (genes ou amostras). O presente trabalho possui como objetivo comparar diferentes medidas de correlação para as três principais tarefas envolvidas na análise de dados de expressão gênica: agrupamento, seleção de atributos e classificação. Dessa forma, é apresentada nesta dissertação uma visão geral da análise de dados de expressão gênica e das diferentes medidas de correlação consideradas para tal comparação. São apresentados também resultados empíricos obtidos a partir da comparação dos coeficientes de correlação para agrupamento de genes, agrupamento de amostras, seleção de genes para o problema de classificação de amostras e classificação de amostras / The development of microarray technology made possible the expression level measurement of hundreds or even thousands of genes simultaneously for various experimental conditions. The huge amount of available data generated the need for computational methods that allow its analysis in an effcient and automated way. In many of the computational methods employed during gene expression data analysis the choice of a proximity measure is necessary. Among the proximity measures available, correlation coefficients have been widely employed because of their ability to capture similarity trends among the compared numeric sequences (genes or samples). The present work has as objective to compare different correlation measures for the three major tasks involved in the analysis of gene expression data: clustering, feature selection and classification. To this extent, in this dissertation an overview of gene expression data analysis and the different correlation measures considered for this comparison are presented. In the present work are also presented empirical results obtained from the comparison of correlation coefficients for gene clustering, sample clustering, gene selection for sample classification and sample classification

Sistema de reconhecimento de fala tem amplo emprego no universo industrial, no aperfeiçoamento de operações e procedimentos humanos e no setor do entretenimento e recreação. O objetivo específico do trabalho foi conceber e desenvolver um sistema de reconhecimento de voz, capaz de identificar comandos de voz, independentemente do locutor. A finalidade precípua do sistema é controlar movimentos de robôs, com aplicações na indústria e no auxílio de deficientes físicos. Utilizou-se a abordagem da tomada de decisão por meio de uma rede neural treinada com as características distintivas do sinal de fala de 16 locutores. As amostras dos comandos foram coletadas segundo o critério de conveniência (em idade e sexo), a fim de garantir uma maior discriminação entre as características de voz, e assim alcançar a generalização da rede neural utilizada. O préprocessamento consistiu na determinação dos pontos extremos da locução do comando e na filtragem adaptativa de Wiener. Cada comando de fala foi segmentado em 200 janelas, com superposição de 25% . As features utilizadas foram a taxa de cruzamento de zeros, a energia de curto prazo e os coeficientes ceptrais na escala de frequência mel. Os dois primeiros coeficientes da codificação linear preditiva e o seu erro também foram testados. A rede neural empregada como classificador foi um perceptron multicamadas, treinado pelo algoritmo backpropagation. Várias experimentações foram realizadas para a escolha de limiares, valores práticos, features e configurações da rede neural. Os resultados foram considerados muito bons, alcançando uma taxa de acertos de 89,16%, sob as condições de pior caso da amostragem dos comandos. / Systems for speech recognition have widespread use in the industrial universe, in the improvement of human operations and procedures and in the area of entertainment and recreation. The specific objective of this study was to design and develop a voice recognition system, capable of identifying voice commands, regardless of the speaker. The main purpose of the system is to control movement of robots, with applications in industry and in aid of disabled people. We used the approach of decision making, by means of a neural network trained with the distinctive features of the speech of 16 speakers. The samples of the voice commands were collected under the criterion of convenience (age and sex), to ensure a greater discrimination between the voice characteristics and to reach the generalization of the neural network. Preprocessing consisted in the determination of the endpoints of each command signal and in the adaptive Wiener filtering. Each speech command was segmented into 200 windows with overlapping of 25%. The features used were the zero crossing rate, the short-term energy and the mel-frequency ceptral coefficients. The first two coefficients of the linear predictive coding and its error were also tested. The neural network classifier was a multilayer perceptron, trained by the backpropagation algorithm. Several experiments were performed for the choice of thresholds, practical values, features and neural network configurations. Results were considered very good, reaching an acceptance rate of 89,16%, under the `worst case conditions for the sampling of the commands.