Spelling suggestions: "subject:"[een] DYNAMIC MODELING"" "subject:"[enn] DYNAMIC MODELING""
31 |
Modelagem dinâmica de transdutor de pressão piezorresistivo para sistemas hidráulicos através de dados experimentais da resposta à entrada degrau / Dynamic modeling for hydraulic piezoresistive pressure transducer using experimental responses due to step inputsEvaldo Ferezin 10 March 2006 (has links)
Este trabalho apresenta a modelagem da parte dinâmica de dois transdutores de pressão tipo piezorresistivo. Geralmente processos de calibração dinâmica usam bancadas sofisticadas, mas neste trabalho procurou-se um processo simples e de baixo custo. A metodologia baseou-se em aumentar lentamente a pressão, fornecida por um sistema hidráulico básico, até que ocorresse a explosão de um diafragma e, com isso, obteve-se um degrau de pressão, supostamente ideal. Considerado como sistemas lineares, com a resposta obtida verificou-se qual a ordem e o tipo de sistema correspondente à resposta do transdutor. Através dos dados experimentais determinou-se os parâmetros dinâmicos do modelo matemático. Foram obtidos modelos matemáticos lineares de segunda ordem de cada transdutor. As respostas dos modelos mostram ter uma concordância satisfatória quando comparadas aos dados experimentais. / This work presents a dynamic modeling method for two hydraulic piezoresistive pressure transducers. Usually the dynamic calibration processes apply sophisticated devices, but in this work a simple and low cost process was searched. The methodology was based on increasing the pressure slowly, supplied by a basic hydraulic system, up to a diaphragm explosion, assumed as an ideal step pressure variation. Considered as linear systems, the responses obtained were verified and compared with a second order linear system. Using the experimental data, the dynamic parameters of the mathematical model were determined for each of the two transducers. The theoretical responses of the models showed to have a satisfactory agreement when compared with the experimental data.
|
32 |
Avaliação da fragmentação florestal no município de Caxias do Sul, RS: implicações ecológicas e proposição de cenários futuros / Assessment of forest fragmentation in the municipality of Caxias do Sul-RS: ecological implications and future scenarios for propositionCemin, Gisele January 2014 (has links)
Na atualidade, as atividades humanas têm levado a conversão das áreas de florestas tropicais em ecossistemas agropastoris e urbanos. Neste processo, a floresta contínua é interrompida por estas barreiras antrópicas as quais são capazes de impedir o fluxo gênico, aumentando a taxa de endocruzamento, o que pode acarretar a diminuição da variabilidade genética e a consequente perda da biodiversidade. A Mata Atlântica, alvo deste estudo, perdeu aproximadamente 71% de sua área original, sendo que seus remanescentes apresentam, na sua grande maioria, áreas pequenas e isoladas uma das outras. Neste contexto, o objetivo desta tese foi de avaliar a evolução da fragmentação florestal no município de Caxias do Sul-RS entre 1985 e 2011, calibrar um modelo dinâmico espacial deste processo e simular um cenário futuro para o ano de 2021, buscando verificar se as mudanças ocorridas ao longo deste período refletem em uma modificação na qualidade ambiental destes fragmentos florestais levando em consideração índices de ecologia de paisagem. Para atingir o objetivo proposto, foi utilizado uma série temporal de imagens do satélite Landsat 5, informações pedológicas, do relevo (hipsometria e clinografia) e índices de ecologia de paisagem. Os resultados indicaram um aumento de aproximadamente 203km² das áreas florestais entre os anos de 1985 e 2011, correspondendo a um incremento 36% das Florestas Estacional Decidual e Ombrófila Mista e também de 20% para o cenário simulado (2021). Em contraponto, houve a diminuição das áreas dos campos de nativos na ordem de 33% de 1985 a 2011 e um pequeno incremento de aproximadamente 5 km² para o cenário simulado (2021). Os índices de ecologia de paisagem indicaram uma melhoria na qualidade ambiental no que tange aos fragmentos florestais, uma vez que tamanho médio e a área interior dos fragmentos aumentaram, ocasionando uma maior coesão entre as áreas florestais vizinhas. Estas informações compõem um indicativo de que a regulamentação do manejo das áreas de Mata Atlântica pode ter favorecido a expansão florestal. Os dados gerados poderão ser utilizados como ferramenta balizadora para a proposição de políticas públicas de visam a consonância entre o desenvolvimento socioeconômico e a preservação ambiental. / At present, human activities have led to conversion of tropical forests in agropastoral and urban ecosystems. In this process, the continuous forest is interrupted by these anthropogenic barriers which are able to prevent gene flow, increasing inbreeding rate, which can lead to decreased genetic variability and the resulting loss of biodiversity. The Atlantic Forest, target of this study, lost approximately 71% of its original area, and its remnants have, for the most part, small areas and isolated from each other. In this context, the aim of this study was to assess the evolution of forest fragmentation in the city of Caxias do Sul-RS between 1985 and 2011, calibrate a spatial dynamic model of this process and simulate a scenario for the year 2021, seeking to verify if the changes and expected present and future scenario reflected in a change in environmental quality of these forest fragments considering landscape ecology indices. To achieve this purpose, we used a time series of satellite images Landsat 5, soil information, topographic (hypsometric and slope) and landscape ecology indices. The results showed an increase of approximately 203km² of forest areas between the years 1985 and 2011, corresponding to an increase of around 36% of Seasonally Deciduous Forests and Mixed Ombrophilous and also 20% for the simulated scenario (2021) In contrast, there was the decrease in areas of native grasslands of about 33% from 1985 to 2011 and a small increase of approximately 5 km² to the simulated scenario (2021). The landscape ecology indices indicated improved environmental quality in terms forest fragments, since average size and the inner area of the fragments increased, leading to greater cohesion between neighboring forest areas. This information generates an indication that the management of the Atlantic Forest may have favored forest expansion. The data generated can be used as guiding tool for proposing public policies aimed at the line between socioeconomic development and environmental preservation.
|
33 |
Modelagem dinâmica e controle de robô manipulador de arquitetura paralela assimétrica de três graus de liberdade. / Dynamic modeling and control of an asymmetric parallel manipulator robot of three degrees of freedom.Rynaldo Zanotele Hemerly de Almeida 31 October 2013 (has links)
Este trabalho trata da modelagem dinâmica e do projeto de sistemas de controle para um robô manipulador de arquitetura paralela assimétrica de três graus de liberdade, correspondente a três translações de seu efetuador no espaço tridimensional, concebido para tarefas de pega-e-põe (pick-and-place). Dentre os desenvolvimentos teóricos, procurou-se estender, para toda gama de robôs paralelos e topologicamente assimétricos, a abrangência dos procedimentos aplicados inicialmente a este caso específico. Foram empregados o Método de Lagrange e o Princípio dos Trabalhos Virtuais na obtenção de modelos dinâmicos baseados em hipóteses simplificadoras de corpos rígidos de massas concentradas e de massas distribuídas. Para o projeto de controladores, foram utilizadas as técnicas de torque computado e torque computado estendido. As formulações correspondentes a estas técnicas foram aprimoradas de forma a permitir o cálculo de esforços de controle de modo equivalente tanto nas coordenadas dos atuadores como nas coordenadas do efetuador e em conformidade com os requisitos de resposta dinâmica definidos para o robô. Tais requisitos podem ser inclusive anisotrópicos, o que se julga ser mais apropriado para robôs paralelos assimétricos. Particularmente em relação ao robô analisado, foram avaliados efeitos de simplificação do modelo dinâmico e da discretização do controlador (incluindo discretização com dupla frequência de amostragem) sobre os erros de controle de posição. Percebeu-se que os requisitos de alto desempenho das tarefas de pega-epõe levam o projeto do controlador ao limite de validade da hipótese simplificadora de corpo rígido e da capacidade de processamento do hardware de controle. / This work deals with the dynamic modeling and the design of control systems for an asymmetric parallel manipulator robot with three degrees of freedom, related to three translations of its end-effector in the tridimensional space, conceived for pick-andplace tasks. Among the theoretical developments, one of them was the extension of the proposed procedures, initially applied to this specific case, to the whole class of topologically asymmetric and parallel mechanisms. The Lagrange Method and the Principle of Virtual work were employed in order to build dynamic models based on rigid body assumption with either lumped or distributed mass simplification hypothesis. For the controllers design, computed torque and extended computed torque techniques were used. The formulations of such techniques were elaborated in order to allow the calculation of the control efforts to be executed equivalently both on the actuators coordinates and on the end-effector coordinates, in accordance with the dynamic response requirements defined for the robot. These requirements may be even anisotropic, what is considered to be more suitable for asymmetric parallel robots. Particularly about the analyzed robot, the effects of dynamic model simplification and controller discretization (including double sample rate discretization) on the position control errors were evaluated. It was realized that the high performance requirements for pick-and-place tasks push the controller design to the limit of validity of the rigid body assumption and of the control hardware processing capacity.
|
34 |
Modelagem e simulação de um sistema hidráulico proposto para funcionar como um amortecedor para pequenas amplitudes / Modeling and simulation for a proposed hydraulic system intended to work as a damper for small amplitudesRodrigo de Oliveira Preti 27 September 2007 (has links)
Este trabalho apresenta um modelo matemático linear para um sistema hidráulico proposto para funcionar como um amortecedor para pequenas amplitudes de deslocamento, da ordem de décimos de milímetros. As dimensões para o sistema hidráulico foram adotadas e parâmetros foram determinados. Foram escolhidos dois conjuntos de parâmetros dimensionais e dois valores para o bulk modulus equivalente, o que permitiu fazer o estudo de quatro situações. A verificação das características dinâmicas do sistema foi efetuada utilizando o conceito de função descritiva. Um programa em Matlab/Simulink com um filtro de Fourier foi implementado para obtenção das curvas da resposta em freqüência. A eficiência deste programa computacional foi comprovada através de sua aplicação a sistemas cujas respostas em freqüência são conhecidas. As curvas das respostas em freqüência do sistema hidráulico e do amortecedor ideal foram comparadas. Considerando tolerância de \'+ OU -\' 1 grau para a fase e de \'+ OU -\' 1 db para magnitude, o sistema hidráulico simulado, com bulk modulus equivalente de 100000 psi teve comportamento similar ao amortecedor ideal na faixa de freqüência de 0,1 a 150 rad/seg. No caso do bulk modulus de 30000 psi, a faixa foi estendida, isto é, de 0,1 a 200 rad/seg. / This work presents a linear mathematical model for a proposed hydraulic system intended to work as a damper for small displacement, of the order of decimals of millimeters. The dimensions of the hydraulic system were adopted and parameters were determined. Two sets of the dimensions and two values for the equivalent bulk modulus were chosen and that allowed the study of four situations. The system dynamic properties were verified using the concept of describing function. A Matlab/Simulink program applying the concept of a Fourier filter was carried out in order to obtain the frequency response curves. The efficiency of this program was verified through its application in systems which frequency responses are known. The systems and the ideal damper frequency response curves were compared. Considering tolerance of \'+ OU -\' 1 degree for the phase and \'+ OU -\' 1 db for magnitude, the simulated hydraulic system with an equivalent bulk modulus of 100000 psi showed to be similar to an ideal damper, for the frequency range from 0.1 to 150 rad/sec. For the equivalent bulk modulus of 30000 psi, the frequency range was extended from 0.1 to 200 rad/sec.
|
35 |
Contribution à la modélisation et à la prédiction de défaillances sur les moteurs diesel marins / Contribution on modeling and prediction of failures on marine diesel enginesMoussa-Nahim, Hassan 22 February 2016 (has links)
Ce travail présente un modèle orienté vers le développement d’un simulateur pour un moteur diesel marin. Dans le but d’avoir un simulateur qui répond aux besoins du moteur étudié au niveau de l’efficacité, la rapidité, et la prise en considération des défauts, le modèle étudié est basé sur des modélisations physiques, semi physique, empirique, mathématique et thermodynamique. Le modèle du moteur est divisé en plusieurs sous-modèles chacun présente un système réel, ces systèmes sont : le refroidissement, la lubrification, l’air, l’injection, la combustion et les émissions. Les sous-modèles et les caractéristiques dynamiques du chaque bloc sont conçus en respectant les équations principales du fonctionnement du moteur ainsi que les données expérimentales collectées du banc d’essai du moteur marin diesel fabriqué par la société SIMB sous la référence 6M26SRP1. Ce modèle a été implémenté sur Matlab/Simulink, et la simulation calcule les variables suivantes : pression, température, efficacité, échange de chaleur, angle de vilebrequin, débit du fuel et émissions, et ceci dans les différentes sous-blocs. Le simulateur est utilisé pour montrer la performance du moteur lors de l’occurrence des défauts et peut aider dans l’application des stratégies d’optimisation pour le dimensionnement, de diagnostic et de pronostic. / This work presents a simulator model of a marine diesel engine based on physical, semi-physical, mathematical and thermodynamic equations, allowing fast predictive simulations. The whole engine system is divided into several functional blocs: cooling, lubrication, air, injection, combustion and emissions. The sub-models and dynamic characteristics of individual blocs are established according to engine working principles equations and experimental data collected from a marine diesel engine test bench for SIMB Company under the reference 6M26SRP1. The overall engine system dynamics is expressed as a set of simultaneous algebraic and differential equations using sub-blocs and S-Functions of Matlab/Simulink. The simulation of this model, implemented on Matlab/Simulink has been validated and can be used to obtain engine performance, pressure, temperature, efficiency, heat release, crank angle, fuel rate, emissions at different sub-blocs. The simulator is used to study the engine performance in faulty conditions, and can be used also to assist marine engineers in FDI (fault detection and isolation) as well as designers to predict the behavior of the cooling system, lubrication system, injection system, combustion, emissions, in order to optimize the dimensions of different components.
|
36 |
Modelling and control of actuated lower limb exoskeletons : a mathematical application using central pattern generators and nonlinear feedback control techniques / Modélisation et commande d'un exosquelette pour les membres inférieurs : approche basée sur les oscillateurs non-linéaires et des techniques de commande non-linéairesAjayi, Michael Oluwatosin 15 November 2016 (has links)
Les exosquelettes représentent des systèmes mécaniques portables qui ont reçu une grande attention de la part de la communauté scientifique ces derrières années, vues les possibilités qu'ils offrent.Ces possibilités concernent principalement les fonctions d'assistance et de réhabilitation des personnes en situation de handicape et personnes âgées, dans un objectif de leur permettre de recouvrir leur facultés motrices. Cependant, d'autres possibilités sont concernées comme permettre à des personnes paraplégiques de remarcher ou de permettre des opérations de manipulation excédent les capacités humaines.Pour permettre la réalisation des fonctions offertes par les robots portables, une connaissance fine de la dynamique du système est requise en relation avec les tâches à réaliser par les sujets. Par ailleurs, des approches de commande sûres qui prennent en compte la sécurité des usagers est nécessaire. Dans cet objectif, des techniques de commande bio-inspirées avec des techniques de commande par découplage non-linéaire sont considérées. Les dernières assurent que la loi de commande est stable et bornée en prenant en considération le domaine de saturation des actionneurs alors que les premières ont inspiré la conception de contrôleurs basés sur les oscillateurs locaux non-linéaires (Central Pattern Generators : CPG).Les CPGs sont modélisés par des réseaux de neurones qui peuvent être représentés par un ensemble d'oscillateurs non-linéaires situés dans la moelle épinière, avec des capacités de génération de signaux rythmiques multidimensionnels synchrones pour remplir des fonctions motrices sous le contrôle de simples signaux de commande. Ces signaux sont supposés être de nature périodique ou semi-périodique, dont la génération pour étudier les systèmes de locmotion humain reste un problème de recherche d'actualité.Dans la présente thèse, l'analyse, la simulation et la commandes des articulations d'un robot portable utilisé pour les membres inférieurs en utilisant, d'un côté les oscillateurs locaux non-linéaires et d'un autre côté des techniques de commande par découplage non-linéaire sont proposés, avec comme objectif final de permtre la mise en œuvre des approches proposées sur la plate-forme expérimentale développée au sein du FSATI (French South African Institute fo Technology).Pour atteindre l'objectif qui a été fixé par les travaux de recherche engagés, une étude de l'état de l'art sur les aspects liés à la connaissance de l'anatomie, la physiologie et l'analyse biomécanique de la marche humaine a été effectuée. Par ailleurs, une étude détaillée des oscillateurs locaux non-linéaires en parallèle avec les approches de commande directe et inverse, ont permis la proposition de stratégies de commande qui couplent les oscillateurs non-linéaires d'un côté et des techniques de découplage non-linéaire d'un autre côté ont été proposées et validées sur des systèmes de plusieurs degrés de liberté. Des simulations intensives ont été conduites afin de vérifier la capacité d'adaptation temps des approches de commandes mises en œuvre avec l'humain ans la boucle.Les contributions de la présente étude concerne deux approches de commande. La première approche concerne l'intégration d'une approche bio-inspirée, basée sur les oscillateurs locaux non-linéaires et la deuxième approche est basée sur les techniques de commande bornée par découplage non-linéaire / Wearable robotic system has become a well sought after mechanism in the field of biomechatronics engineering due to the the various possibilities it possess. These possibilities encompass the assistive and rehabilitative protocols rendered to disabled and elderly people, in order to enable them regain control of their limbs and of course increase the abilities of able-bodied persons. It therefore clearly drives the motive of bringing back paraplegics back on their feet as well as executing difficult task beyond human ability.Achieving the intended function of wearable robots requires the model dynamics of the physical system in relation to the tasks required to be performed by subjects. This demands a proper control measure which takes into account the safety of the wearer. For this purpose, bio-inspired control techniques and bounded nonlinear feedback controllers are considered. The latter control design ensures that the stipulated power required is not exceeded as well as the saturation of the actuator, while the former motivates the design of controllers based on the concept of Central Pattern Generators (CPG). CPGs are characterised as biological neural networks which can be represented by a set of coupled nonlinear oscillator situated in the spinal cord of mammals, having the capability of generating coordinated multidimensional rhythmic signals for the purpose of locomotion, under the control of simple input signals. These rhythmic signals are termed to be periodic or quasi-periodic in nature, hence performing this task in robotics and animal motor control has been a perpetual research problem. The movement of the lower limb of humans thus present a platform to investigate and address this difficulty.In this thesis, the analysis, simulation, and control of joints which relate to the human lower limbs via CPGs and feedback control techniques are investigated with an aim of practically implementing the control strategies using a lower limb exoskeleton is presented. To accomplish this goal, it is expedient to have comprehensive knowledge of the anatomy, physiology and the normal gait biomechanics of the human lower limbs. Understanding the theories, principles and mathematical background of nonlinear oscillators are also required. Control strategies using the inverse and the forward dynamics approach based on different types of coupled nonlinear oscillators and nonlinear feedback control techniques were considered for single/multiple degrees of freedom (DoF). Simulations and results were presented to verify the controller-human system ability to constantly and dynamically track and readapt its control parameters to maintain its desired motion dynamics, with reduced control torque values.This work basically deals with two distinct method of control systems; one which integrates bio-inspired methods with classical and nonlinear control techniques to govern the exoskeletons' joints with a human in the loop, and another which utilises bounded nonlinear feedback control techniques for same purpose
|
37 |
Analyse de la stabilité de la coupe d'un procédé d'usinage robotisé / Stability analysis in robotic machining processMoussavi, Said 31 May 2016 (has links)
La productivité des processus d'usinage robotisé est très souvent limitée par le manque de rigidité des robots et les problèmes vibratoires relatifs à l’instabilité de la coupe. L’analyse de l’instabilité de la coupe en usinage robotisé est un problème difficile en raison de la variabilité du comportement dynamique du robot dans son espace de travail. Par conséquent, le phénomène de broutement en usinage robotisé dépend non seulement des paramètres de coupe mais également de la configuration du robot. Le premier objectif de cette thèse est de déterminer une méthode de modélisation dynamique du robot, adaptée du point de vue de l’analyse des vibrations et de la stabilité en usinage robotisé. Cette approche a été réalisée sur le robot d’usinage industriel ABB IRB6660. Une démarche de recalage a été mise en place afin de déterminer les paramètres du modèle numérique du robot. Ensuite, une présentation 3D de la limite de stabilité en usinage robotisé prenant en compte les variations du comportement dynamique du robot est réalisée. Le deuxième objectif consiste à optimiser le procédé vis-à-vis de la stabilité de la coupe. Les variations du comportement dynamique du robot sont exploitées par la gestion des redondances fonctionnelles afin d’optimiser la configuration du robot du point de vue de la stabilité. L’analyse numérique a montré et les essais expérimentaux d’usinage ont confirmé la possibilité de passer de la zone instable à la zone stable par la gestion de la redondance fonctionnelle sans modifier les paramètres de coupe. / Productivity in robotic machining processes can be limited by the low rigidity of the overall structure and vibration instability (chatter). The robot’s dynamic behavior, due to changes in its posture along a machining trajectory, varies within its workspace. Chatter in robotic machining therefore depends not only on the cutting parameters but also on the robot configuration. The first objective of this thesis is to determine a dynamic modeling approach of the robot in order to analyze the vibration and the stability in robotic machining. This modeling approach has been realized to dynamic modeling of an ABB IRB6660 industrial robot. The numerical model parameters are adjusted on the basis of experimental modal identifications. Then, a three-dimensional representation of stability lobes diagram for the prediction to take into account the robot dynamic behavior variations in machining trajectory is established. The second objective is to optimize the robot configurations regarding stability. The dynamic behavior variations of the robot in the workspace are exploited through functional redundancy management in order to optimize robot configurations with respect to machining stability. The numerical analyze demonstrated and experimental machining tests confirmed that stability conditions in machining operations can be achieved by managing functional redundancy without changing the cutting parameters.
|
38 |
Modelagem e validação experimental de uma estrutura flexívelMarcelo Rodolfo de Assis 04 December 2012 (has links)
Regarding the flexible structures and vibration area, this paper aims to develop, implement and validate an analytical and experimental model of cantilevered beam in order to analyze the natural frequencies and vibration modes through a prototype assembled in lab. The analytical model was obtained using the theory of Euler-Bernoulli beams and Lagrange formalism. To decouple the system associated variables was applied the assumed method. The analytical model implementation and simulation were developed using the MATLAB software. The data acquisition process was developed by LABVIEW software. Complementary tests regarding shaker and servo motors were performed and showed in order to aggregate for the experimental
procedure. The experimental and analytical results of this paper present good approximation between them allowing the analytical model of cantilevered beam characterization. / Esta dissertação tem como proposta desenvolver, implementar e validar o modelo matemático e experimental de uma viga flexível simplesmente engastada, permitindo analisar as frequências naturais e modos de vibração, através de um protótipo montado em laboratório. O modelo analítico foi obtido utilizando-se a teoria de vigas de Euler-Bernoulli e o formalismo de Lagrange. Para discretização das variáveis acopladas ao sistema, foi utilizado o Método dos Modos Assumidos. Para a implementação do modelo analítico e a simulação dos resultados foi utilizado o software MATLAB. O processo de aquisição de dados experimentais foi
desenvolvido utilizando o software Labview. Testes complementares relacionados à utilização de shaker e servo motor também são apresentados. Os resultados experimentais e analíticos apresentam boa aproximação entre si, permitindo validar o modelo analítico da viga em estudo.
|
39 |
Vertical Distribution of Daily Migrating Mesopelagic Fish in Respect to Nocturnal LightsPrihartato, Perdana 12 1900 (has links)
The nighttime distribution of vertically migrating mesopelagic fish in relation to
nocturnal light was studied during a circumglobal survey, in the Red Sea, and in a fjord at
high latitude. The study was based on data derived from ship borne echo sounders
(circumglobal and the Red Sea) as well as using upward looking echo sounders mounted
on the bottom (Masfjorden, Norway). We also applied a numerical model for analyzing
diel vertical migration patterns. The effect of the lunar cycle was the focus in studies at
low latitudes, while seasonal changes in nocturnal light climate was in focus at high
latitude. Lunar phase significantly affected the distribution of mesopelagic fish at the
global scale and in the Red Sea. During nights near full moon, scattering layers of
mesopelagic fish distributed deeper than during darker phases of the moon. At high
latitude, mesopelagic fish switched its behavior along with seasonal changes in nocturnal
lights. In autumn, the population of the studied fish (Maurolicus mueleri) formed
separated layers. Juveniles performed normal diel vertical migration followed by
midnight sinking, with midnight sinking mainly related to temperature minima and also
for avoiding predators. Meanwhile the adults did not migrate vertically, reducing
foraging but increasing the adult survival. From late winter to mid-Spring, interrupted
ascents behavior was noted in the afternoon. Predator avoidance, satiation, and finding
temperature optimum might be the reason behind interrupted ascents. At lighter nights in
mid-summer, M. muelleri took on schooling behavior, likely as an anti-predator behavior
permitting access to the upper waters in the absence of darkness.
|
40 |
Accounting for Individual Speaker Properties in Automatic Speech RecognitionElenius, Daniel January 2010 (has links)
In this work, speaker characteristic modeling has been applied in the fields of automatic speech recognition (ASR) and automatic speaker verification (ASV). In ASR, a key problem is that acoustic mismatch between training and test conditions degrade classification per- formance. In this work, a child exemplifies a speaker not represented in training data and methods to reduce the spectral mismatch are devised and evaluated. To reduce the acoustic mismatch, predictive modeling based on spectral speech transformation is applied. Follow- ing this approach, a model suitable for a target speaker, not well represented in the training data, is estimated and synthesized by applying vocal tract predictive modeling (VTPM). In this thesis, the traditional static modeling on the utterance level is extended to dynamic modeling. This is accomplished by operating also on sub-utterance units, such as phonemes, phone-realizations, sub-phone realizations and sound frames. Initial experiments shows that adaptation of an acoustic model trained on adult speech significantly reduced the word error rate of ASR for children, but not to the level of a model trained on children’s speech. Multi-speaker-group training provided an acoustic model that performed recognition for both adults and children within the same model at almost the same accuracy as speaker-group dedicated models, with no added model complexity. In the analysis of the cause of errors, body height of the child was shown to be correlated to word error rate. A further result is that the computationally demanding iterative recognition process in standard VTLN can be replaced by synthetically extending the vocal tract length distribution in the training data. A multi-warp model is trained on the extended data and recognition is performed in a single pass. The accuracy is similar to that of the standard technique. A concluding experiment in ASR shows that the word error rate can be reduced by ex- tending a static vocal tract length compensation parameter into a temporal parameter track. A key component to reach this improvement was provided by a novel joint two-level opti- mization process. In the process, the track was determined as a composition of a static and a dynamic component, which were simultaneously optimized on the utterance and sub- utterance level respectively. This had the principal advantage of limiting the modulation am- plitude of the track to what is realistic for an individual speaker. The recognition error rate was reduced by 10% relative compared with that of a standard utterance-specific estimation technique. The techniques devised and evaluated can also be applied to other speaker characteristic properties, which exhibit a dynamic nature. An excursion into ASV led to the proposal of a statistical speaker population model. The model represents an alternative approach for determining the reject/accept threshold in an ASV system instead of the commonly used direct estimation on a set of client and impos- tor utterances. This is especially valuable in applications where a low false reject or false ac- cept rate is required. In these cases, the number of errors is often too few to estimate a reli- able threshold using the direct method. The results are encouraging but need to be verified on a larger database. / QC 20110502 / Pf-Star / KOBRA
|
Page generated in 0.0629 seconds