Orientadores: Wagner Caradori do Amaral, Ricardo Jose Gabrielli Barreto Campello / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-04T02:57:58Z (GMT). No. of bitstreams: 1 Rosa_Alexda_M.pdf: 896715 bytes, checksum: 1baf3dbaef2a1280f09feabed84d996c (MD5) Previous issue date: 2005 / Resumo: Este trabalho analisa a modelagem de sistemas nao-lineares utilizando modelos de Wiener/Volterra com funcoes ortonormais de Kautz. Os modelos de Volterra sao uma generalizacao do modelo resposta ao impulso para a descricao de sistemas naolineares. Esses modelos necessitam de um numero consideravel de termos para a representacao dos kernels de Volterra. Essa complexidade pode ser reduzida utilizando-se uma representacao do tipo Wiener/Volterra, em que os kernels sao desenvolvidos utilizando uma base de funcoes ortonormais. Sao discutidos aspectos da selecao dos parametros livres (polos) que caracterizam essas funcoes, particularmente a selecao otima dos polos complexos das funcoes de Kautz. Este problema e resolvido minimizando-se o limitante superior do erro que surge a partir da aproximação truncada dos kernels de Volterra usando-se as funcoes de Kautz. Obtem-se a solu¸cao analitica para a escolha otima de um dos parametros relacionados com o polo de Kautz, sendo os resultados validos para modelos Wiener/Volterra de qualquer ordem. Apresentam-se ainda resultados de simulacoes que ilustram a metodologia apresentada, bem como a modelagem de um sistema de levitacao magnetica / Abstract: This work investigates the modelling of nonlinear systems using the Wiener/Volterra models with Kautz orthonormal functions. The Volterra models constitute a generalization of the impulse response model to describe nonlinear systems. Such models require a large number of terms for representing the Volterra kernels. However, this complexity can be reduced by using Wiener/Volterra models, in which the kernels are expanded using an orthonormal basis functions. Aspects about selection of the free parameters (poles) characterizing theses functions are discussed, in particular the optimal selection of the complex poles of the Kautz functions. This problem is solved by minimizing the upper bound of the error arising from the truncated approximation of Volterra kernels using Kautz functions. An analytical solution for the optimal choice of one of the parameters related to the Kautz pole is thus obtained, with the results valid for any-order Wiener/Volterra models. Simulations that illustrate the methodology described above are presented. Also, the modelling of a magnetic levitation system is discussed. / Mestrado / Engenharia / Mestre em Engenharia Elétrica

In this Master's thesis, Maximum Likelihood-based parametric identification methods for discrete-time SISO Hammerstein models from perturbed observations on both input and output, are investigated. Hammerstein models, consisting of a static nonlinear block followed by a dynamic linear one, are widely applied to modeling nonlinear dynamic systems, i.e., dynamic systems having nonlinearity at its input. Two identification methods are proposed. The first one assumes a Hammerstein model where the input signal is noise-free and the output signal is perturbed with colored noise. The second assumes, however, white noises added to the input and output of the nonlinearity and to the output of the whole considered Hammerstein model. Both methods operate directly in the time domain and their properties are illustrated by a number of simulated examples. It should be observed that attention is focused on derivation, numerical calculation, and simulation corresponding to the first identification method mentioned above.

Ce rapport porte sur le sujet de recherche de l'identification boîte noire du système non linéaire. En effet, parmi toutes les techniques nombreuses et variées développées dans ce domaine de la recherche ces dernières décennies, il semble toujours intéressant d'étudier l'approche réseau de neurones dans l'estimation de modèle de système complexe. Même si des modèles précis ont été obtenus, les principaux inconvénients de ces techniques restent le grand nombre de paramètres nécessaires et, en conséquence, le coût important de calcul nécessaire pour obtenir le niveau de pratique de la précision du modèle désiré. Par conséquent, motivés pour remédier à ces inconvénients, nous avons atteint une méthodologie complète et efficace du système d'identification offrant une précision équilibrée, la complexité et les modèles de coûts en proposant, d'une part, de nouvelles structures de réseaux de neurones particulièrement adapté à une utilisation très large en matière de modélisation système pratique non linéaire, d'autre part, un simple et efficace technique de réduction de modèle, et, troisièmement, une procédure de réduction de coût de calcul. Il est important de noter que ces deux dernières techniques de réduction peut être appliquée à une très large gamme d'architectures de réseaux de neurones sous deux simples hypothèses spécifiques qui ne sont pas du tout contraignant. Enfin, la dernière contribution importante de ce travail est d'avoir montré que cette phase d'estimation peut être obtenue dans un cadre robuste si la qualité des données d'identification qu'il oblige. Afin de valider la procédure d'identification système proposé, des exemples d'applications entraînées en simulation et sur un procédé réel, de manière satisfaisante validé toutes les contributions de cette thèse, confirmant tout l'intérêt de ce travail. / This report concerns the research topic of black box nonlinear system identification. In effect, among all the various and numerous techniques developed in this field of research these last decades, it seems still interesting to investigate the neural network approach in complex system model estimation. Even if accurate models have been derived, the main drawbacks of these techniques remain the large number of parameters required and, as a consequence, the important computational cost necessary to obtain the convenient level of the model accuracy desired. Hence, motivated to address these drawbacks, we achieved a complete and efficient system identification methodology providing balanced accuracy, complexity and cost models by proposing, firstly, new neural network structures particularly adapted to a very wide use in practical nonlinear system modeling, secondly, a simple and efficient model reduction technique, and, thirdly, a computational cost reduction procedure. It is important to notice that these last two reduction techniques can be applied to a very large range of neural network architectures under two simple specific assumptions which are not at all restricting. Finally, the last important contribution of this work is to have shown that this estimation phase can be achieved in a robust framework if the quality of identification data compels it. In order to validate the proposed system identification procedure, application examples driven in simulation and on a real process, satisfactorily validated all the contributions of this thesis, confirming all the interest of this work.

Stricter emission legislation for heavy trucks in combination with the customers demand on low fuel consumption has resulted in intensive technical development of engines and their control systems. To control all these new solutions it is desirable to have reliable models for important control variables. One of them is the air mass-flow, which is important when controlling the amount of recirculated exhaust gases in the EGR system and to make sure that the air to fuel ratio is correct in the cylinders. The purpose with this thesis was to use system identification theory to develop a model for the air mass-flow through the compressor. First linear black-box models were developed without any knowledge of the physics behind. The collected data was preprocessed to work in the modeling procedure and then models with one or more inputs where built according to the ARX model structure. To further improve the models performance, non-linear regressors was developed from physical relations for the air mass-flow and used to form grey-box models of the air mass-flow.In conclusion, the performance was evaluated through comparing the estimated air mass-flow from the best model with the estimate that an extended Kalman filter together with a physical model produced. / Hårdare utsläppskrav för tunga lastbilar i kombination med kundernas efterfrågan på låg bränsleförbrukning har resulterat i en intensiv utveckling av motorer och deras kontrollsystem. För att kunna styra alla dessa nya lösningar är det nödvändigt att ha tillförlitliga modeller över viktiga kontrollvariabler. En av dessa är luftmassflödet som är viktig när man ska kontrollera den mängd avgaser som återcirkuleras i EGR-systemet och för att se till att kvoten mellan luft och bränsle är korrekt i motorns cylindrar. Syftet med det här examensarbetet var att använda systemidentifiering för att ta fram en modell över luftmassflödet förbi kompressorn. Först togs linjära svartboxmodeller fram utan att ta med någon kunskap om den bakomliggande fysiken. Insamlade data förbehandlades för att passa in i modelleringsproceduren och efter det skapades i enlighet med ARX-modellstrukturen modeller med en eller flera insignaler. För att ytterligare förbättra modellernas prestanda togs icke-linjära regressorer fram med hjälp av fysikaliska relationer för luftmassflödet. Dessa användes sedan för att skapa gråboxmodeller av luftmassflödet. Avslutningsvis utvärderades prestandan genom att det estimerade luftmassflödet från den bästa modellen jämfördes med det estimat som ett utökat kalmanfilter tillsammans med fysikaliska ekvationer genererade.

This thesis presents System Identification based On-Line Health Monitoring to analyse the dynamic behaviour of the Switch-Mode Power Converter (SMPC), detect, and diagnose anomalies in passive electronic components. The anomaly detection in this research is determined by examining the change in passive component values due to degradation. Degradation, which is a long-term process, however, is characterised by inserting different component values in the power converter. The novel health-monitoring capability enables accurate detection of passive electronic components despite component variations and uncertainties and is valid for different topologies of the switch-mode power converter. The need for a novel on-line health-monitoring capability is driven by the need to improve unscheduled in-service, logistics, and engineering costs, including the requirement of Integrated Vehicle Health Management (IVHM) for electronic systems and components. The detection and diagnosis of degradations and failures within power converters is of great importance for aircraft electronic manufacturers, such as Thales, where component failures result in equipment downtime and large maintenance costs. The fact that existing techniques, including built-in-self test, use of dedicated sensors, physics-of-failure, and data-driven based health-monitoring, have yet to deliver extensive application in IVHM, provides the motivation for this research ... [cont.].

The objective of this thesis is to develop algorithms for application-oriented input design. This procedure takes the model application into account when designing experiments for system identification. This thesis is divided into two parts. The first part considers the theory of application-oriented input design, with special attention to Model Predictive Control (MPC). We start by studying how to find a convex approximation of the set of models that result in acceptable control performance using analytical methods when controllers with no closed-form control law, for e.g., MPC are employed. The application-oriented input design is formulated in time domain to enable handling of signals constraints. The framework is extended to closed-loop systems where two cases are considered i.e., when the plant is controlled by a general but known controller and for the case of MPC. To this end, an external stationary signal is designed via graph theory. Different sources of uncertainty in application-oriented input design are investigated and a robust application-oriented input design framework is proposed. The second part of this thesis is devoted to the problem of estimating the number of occupants based on the information available to HVAC systems in buildings. The occupancy estimation is first formulated as a two-tier problem. In the first tier, the room dynamic is identified using temporary measurements of occupancy. In the second tier, the identified model is employed to formulate the problem as a fused-lasso problem. The proposed method is further developed to be used as a multi-room estimator using a physics-based model. However, since it is not always possible to collect measurements of occupancy, we proceed by proposing a blind identification algorithm which estimates the room dynamic and occupancy, simultaneously. Finally, the application-oriented input design framework is employed to collect data that is informative enough for occupancy estimation purposes. / <p>QC 20170620</p>

No description available.

Orientador: Robson Pederiva / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Mecânica / Made available in DSpace on 2018-08-28T09:02:45Z (GMT). No. of bitstreams: 1 Sanches_FabioDalmazzo_D.pdf: 3111854 bytes, checksum: 7a85cd92fff24bc0ca9170821fb61025 (MD5) Previous issue date: 2015 / Resumo: Este trabalho aborda a identificação simultânea do desbalanceamento e empeno de eixo em rotores a partir do modelo matemático tanto do rotor como das falhas estudadas. Esses dois fenômenos são síncronos à rotação do rotor e, por isso, difíceis de serem separados um do outro. A partir do modelo do rotor representado na forma de espaço de estados, é possível aplicar a definição de matrizes de correlações de modo a gerar um algoritmo que correlaciona os parâmetros de falhas com as respostas do sistema, possibilitando que as falhas sejam identificadas tanto em magnitude como em localização. O estimador é baseado na equação matricial de Lyapunov. Como o número de respostas medidas é inferior ao total de graus de liberdade do rotor, um sistema auxiliar (filtro) é necessário para gerar correlações adicionais que possibilitem a identificação das falhas em estudo. Transformações de coordenadas através da matriz de observabilidade são necessárias para descrever o sistema através das poucas respostas medidas do rotor. Como um modelo matemático confiável é essencial para o sucesso da identificação, grandezas desconhecidas tais como: parâmetros de mancal, rigidez angular do acoplamento e amortecimento do sistema foram identificados através de otimização pelo método Evolução Diferencial, comparando-se as FRF's medidas e simuladas. Técnicas de redução de ordem do modelo ajustado do rotor foram empregadas de modo a reduzir o esforço computacional na determinação da matriz de observabilidade, possibilitando o uso de um filtro de ordem reduzida. Os estudos ocorrem de forma numérica e experimental para dois tipos de rotores: Laval e dois discos. Simulações mostram como o empeno de eixo altera a dinâmica dos rotores quando comparadas com as respostas ao desbalanceamento puro. Na presença do empeno, várias configurações de desbalanceamento foram simuladas e posteriormente introduzidas nas bancadas experimentais. O algoritmo de identificação se mostrou robusto na caracterização dessas duas falhas e o método contribui com o estado da arte na área de detecção de falhas / Abstract: This work is about simultaneous unbalance and shaft bow identification in rotor using the mathematical model of both the rotor and the studied failures. These two phenomena are synchronous with the rotation of the rotor, causing difficulties in separating them from each other. From the rotor model represented in state space form, it is possible to use the definition of correlation matrix in order to generate an algorithm that correlates the fault parameters with the system responses, enabling the faults to be identified in amplitude and phase. The estimator comes form the Lyapunov matrix equation. As the number of measured available responses are lower than total number of rotor degrees of freedom, an auxiliary system (filter) is required to generate additional correlations that enable the studied faults identification. Coordinate transformations though the observability matrix are necessary to describe the system by means of some few measured outputs. As a reliable mathematical model is essential for the success of the identification process, unknown parameters such as: bearing coefficients, coupling angular stiffness and damping systems were identified using the Differential Evolution optimization technique in which the experimental and simulated FRF's were compared and adjusted. Model order reduction techniques were used on the updated rotor model to reduce computational costs in determining the observability matrix, which allows the usage of a lower filter order. The studies were performed numerically and experimentally for two different test rigs: Laval rotor and two disc rotor. Simulations show the shaft bow changes the dynamic behavior of the rotor when compared to the pure unbalanced rotor responses. In the presence of bow, various unbalance configurations were simulated and thereafter introduced in the test rigs. The identification algorithm showed robustness in identifying these two faults and this method contributes with the state of the art in the field of faults identification / Doutorado / Mecanica dos Sólidos e Projeto Mecanico / Doutor em Engenharia Mecânica

Passive network reconstruction is the process of learning a structured (networked) representation of a dynamic system through the use of known information about the structure of the system as well as data collected by observing the inputs into a system along with the resultant outputs. This work demonstrates an improvement on an existing network reconstruction algorithm so that the algorithm is capable of consistently and perfectly reconstructing a network when system inputs and outputs are measured without error. This work then extends the improved network reconstruction algorithm so that it functions even in the presence of noise as well as the situation where inputs into the system are unknown. Furthermore, this work demonstrates the capability of the new extended algorithms by reconstructing financial networks from stock market data, and then performing an analysis to understand the vulnerabilities of the reconstructed network to destabilization through localized attacks. The creation of these improved and extended algorithms has opened many theoretical questions, paving the way for future research into network reconstruction.

This thesis deals with automated tuning of drive controller. To achieve this goal, system identification is needed. Therefore, the issue of identification is described at the beginning of this thesis. Spectral analysis was selected from many described methods. It was implemented in Matlab and also in C# language where was used averaging and Blackman-Tukey method. The C# application is linked to Beckhoff TwinCAT 3 and TwinCAT 2 runtime sytems that enable connections with real drive. Next, the problem of drive regulation is discussed and the results of using spectral analysis on real drives are shown. At the end of the thesis is described the algorithm of setting the speed controller with different types of frequency converters.