Nos sistemas de distribuição de alta e média tensão tem-se aumentado a utilização de linhas de distribuição de energia subterrâneas ou cabos subterrâneos. A ocorrência de faltas nas linhas afeta negativamente a qualidade da energia e o correto funcionamento da rede. O processo que leva a uma falta nos cabos é gradual e está caracterizado por uma série de subciclos de faltas incipientes associadas a uma tensão de arco. Estas, muitas vezes, passam despercebidas e, eventualmente, resultam em uma falta permanente. Os métodos clássicos de localização de faltas como as metodologias baseadas no cálculo da impedância aparente, as baseadas na inteligência artificial e as baseadas nas ondas viajantes são, habitualmente, aplicadas ao sistema depois de uma falta permanente acontecer e precisam de um ou mais ciclos do sinal para entregar uma resposta razoável. No entanto, as faltas nos cabos são um processo gradual, de curta duração (entre ¼ e ½ ciclo do sinal) e seria desejável localizar a falta incipiente antes de tornar-se permanente. Nesse contexto, esta pesquisa aborda o problema de localização de faltas incipientes. Assim, nesta tese propõe-se uma nova técnica de localização de faltas incipientes usando medições em um terminal, no domínio do tempo e que utiliza componentes de fase. Desta forma, são desenvolvidas duas novas formulações do modelo elétrico do sistema de distribuição com cabos subterrâneos durante uma falta incipiente. A abordagem proposta considera simultaneamente na sua formulação características da falta incipiente e dos sistemas de distribuição de energia, como a tensão de arco, o modelo Π nominal de parâmetros concentrados do cabo subterrâneo, o desequilíbrio do sistema e a condição da carga. A estimativa da distância da falta, junto com os parâmetros da falta incipiente, é obtida a partir da solução de um sistema sobredeterminado de equações lineares pela aplicação do método de mínimos quadrados ponderados não negativos. As formulações propostas permitem estimar a distância da falta em termos da reatância da linha até a falta. Além disso, é proposto um processo de compensação de corrente para estimar a corrente de falta; é aplicado um pré-processamento dos dados de entrada para suavizar o efeito do ruído que pode conter o sinal e, é aplicado um pós-processamento dos resultados para refinar e entregar a melhor estimativa obtida durante o processo de localização da falta. O desempenho da técnica proposta é avaliado mediante estudos de casos simulados em um circuito real de distribuição no Alternative Transients Program (ATP/EMTP) considerando análises de sensibilidade e comparativa. Também, o modelo da falta incipiente foi programado utilizando a ferramenta de MODELS do ATP/EMTP. Os resultados obtidos, considerando faltas incipientes simuladas que avaliam a influência da variação da magnitude de tensão de arco, do ruído aleatório inserido na tensão de arco, da distância da falta, da taxa de amostragem, do carregamento do sistema, do modelo de tensão de arco e de incertezas nas medições, indicam claramente que a abordagem proposta possui validade como técnica de localização de faltas incipientes, apresentando erros médios globais de 1,60% e 0,93%, respectivamente para cada formulação proposta. / The use of underground power distribution lines or underground cables in the high and medium voltage distribution systems has increased dramatically in recent years. The fault occurrence in the distribution lines negatively affects the power quality and the correct network operation. The process which leads to a fault in underground cables is gradual and is characterized by a series of sub-cycles of incipient faults associated with an arc voltage. These often are unnoticed and, eventually, results in a permanent fault. Classical fault localization methods such as the based-impedance, the based on artificial intelligent and the based on traveling waves are, usually, applied to the system after a permanent fault occurrence and need one or more signal cycles for providing a reasonable response. However, the faults in cables are a gradual process, with short duration (between ¼ to ½ of signal cycle) and would be desirable to locate the fault before this becomes a permanent fault. In this context, this research approaches the incipient faults location problem. Thus, in this thesis is proposed a new incipient fault location technique using single-end terminal measurement, in time-domain and employing phase components. In this way, two new formulations of the electrical model of the distribution system with underground cables during an incipient fault are developed. The proposed approach considers simultaneously in its formulation, incipient fault type and power distribution systems characteristics as arc voltage, unbalanced operation, load conditions and complete line model. The fault distance estimation, together with the incipient fault parameters, it is obtained from the solution of an overdetermined linear system of equations by the application of the non-negative weighted least squares estimator method. The proposed formulations allow estimating the fault distance in terms of the line reactance up to the fault. In addition, a load current compensation strategy is proposed to reduce its effect in the fault current estimation; an input data pre-processing is applied to smooth out the noise effect and a post-processing of the results is performed for estimation refinement and to provide the best estimate obtained during the fault location process. The proposed technique performance is evaluated through simulated cases studies in a real-life distribution network with underground cable data in the Alternative Transients Program (ATP/EMTP) considering sensitivity and comparative analyzes. Also, the fault model was programmed using the MODELS tool of ATP/EMTP. The obtained results, considering simulated incipient faults, which evaluate the influence of variations in the arc voltage magnitude, random noise percentage inserted in the arc voltage, fault distance, sampling rate, load dynamics, the arc voltage model and uncertainties in measurements, indicate clearly that the proposed approach is valid as incipient faults location technique, showing overall average errors of 1,60% and 0,93%, respectively for each proposed formulation.

Submitted by Patrícia Cerveira (pcerveira1@gmail.com) on 2017-05-29T17:39:15Z No. of bitstreams: 1 Daniel Souto Dissertação.pdf: 2907570 bytes, checksum: 0379fd854c367debb9d8b6237a881084 (MD5) / Rejected by Biblioteca da Escola de Engenharia (bee@ndc.uff.br), reason: Bom dia, Patrícia! Rejeitei este item, pois fiquei com dúvida se é realmente acesso restrito. Conforme orientações da Jane, o acesso é aberto. Você está colocando acesso restrito. O acesso somente é restrito quando a tese ou dissertação serão publicadas e se exige ineditismo, ou quando o produto do trabalho tem segredo empresarial ou virará patente. Nos demais casos, é aberto. Aguardo retorno. Atenciosamente, Catarina Ribeiro Bibliotecária BEE - Ramal 5992 on 2017-06-29T13:54:55Z (GMT) / Submitted by Patrícia Cerveira (pcerveira1@gmail.com) on 2017-06-29T18:56:23Z No. of bitstreams: 1 Daniel Souto Dissertação.pdf: 2907570 bytes, checksum: 0379fd854c367debb9d8b6237a881084 (MD5) / Approved for entry into archive by Biblioteca da Escola de Engenharia (bee@ndc.uff.br) on 2017-07-03T12:39:19Z (GMT) No. of bitstreams: 1 Daniel Souto Dissertação.pdf: 2907570 bytes, checksum: 0379fd854c367debb9d8b6237a881084 (MD5) / Made available in DSpace on 2017-07-03T12:39:19Z (GMT). No. of bitstreams: 1 Daniel Souto Dissertação.pdf: 2907570 bytes, checksum: 0379fd854c367debb9d8b6237a881084 (MD5) / O problema de diagnóstico de faltas em linhas de transmissão constitui um dos principais desafios para gestão técnica de instalações de transmissão. A assertividade nesta atividade é fundamental para suporte à tomada de decisão, reduzindo as taxas de indisponibilidade e promovendo o restabelecimento célere da função transmissão, contribuindo para melhoria da qualidade do serviço e reduzindo os impactos financeiros advindos de reduções na parcela variável. Este documento apresenta uma proposta de sistema inteligente para classificação e localização de faltas em linhas de transmissão. Os algoritmos utilizados são baseados nos chamados modelos neurais autônomos, que incluem técnicas analíticas para seleção de entradas e especificação automática da estrutura sem a necessidade do uso de um conjunto independente de dados para validação. Ao utilizar a inferência bayesiana para especificação e treinamento de perceptrons de múltiplas camadas (MLPs), o sistema inteligente fornece respostas probabilísticas para classificação do tipo de defeito e também para a distância da falta em relação à subestação monitorada. Para desenvolvimento dos modelos são utilizados dados técnicos de uma linha de transmissão integrante do Sistema Interligado Nacional (SIN), a qual é modelada em um “software” de simulação de transitórios eletromagnéticos, ATP, visando estabelecer os diversos cenários de falta. Foram analisados dois tipos de rede equivalente, uma detalhada e outra simples, de forma a precisar qual o melhor modelo e se há diferenças significativas nos resultados em termos de representação das faltas. As bases de dados com as oscilografias de tensão e corrente obtidas para cada tipo de defeito são utilizadas para treinamento e teste do sistema inteligente, sendo demonstrando o potencial dos algoritmos utilizados. / The problem of fault diagnosis in transmission lines is one of the main challenges for the technical management of transmission facilities. The assertiveness on this activity is crucial to support decision making, reducing unavailability rates and promoting rapid reinstatement of the transmission function, contributing to the improvement of service quality and reducing the financial impacts arising from reductions in the variable portion. This document presents a proposal of intelligent system for classification and location of faults in transmission lines. The algorithms used are based on the so-called autonomous neural models which include analytical techniques for input selection and automatic structure specification without the need for an independent set of data for validation. Using Bayesian inference for specification and training of multilayer perceptrons (MLPs), the intelligent system provides probabilistic responses for classification of the type of fault and also for the distance of the fault from the monitored substation. Thus for the development of the models, technical data are used of a transmission line that is part of the National Interconnected System (SIN) which is modeled in an electromagnetic transient simulation software, ATP, aiming to establish the various fault scenarios. Furthermore, two types of equivalent network were analyzed, one detailed and one simple, in order to specify the best model and if there were significant differences in results in terms of fault representation. The databases with voltage and current oscillographs obtained for each type of fault are used for training and testing of the intelligent system, demonstrating the potential of the algorithms used.

Les travaux présentés dans ce mémoire traitent de la synthèse d'algorithmes de diagnostic de défauts simples et multiples. L'objectif vise à proposer une stratégie de diagnostic à minimum de redondance analytique en exploitant au mieux les informations de redondance matérielle éventuellement disponibles sur le système. Les développements proposés s'inscrivent dans une démarche de coopération et d'agrégation des méthodes de diagnostic et la construction optimale d'un diagnostic global en fonction des diagnostiqueurs locaux. Les travaux réalisés se veulent génériques dans le sens où ils mêlent à la fois les concepts et outils de deux communautés : ceux de la communauté FDI (Fault Detection and Isolation) et ceux de la communauté DX (Diagnosis) dont les bases méthodologiques sont issues des domaines informatiques et intelligence artificielle. Ainsi, le problème de détection (ainsi que le problème de localisation lorsque les contraintes structurelles le permettent) est résolu à l'aide des outils de la communauté FDI tandis que le problème de localisation est résolu à l'aide des concepts de la communauté DX, offrant ainsi une démarche méthodologique agrégée. La démarche méthodologique se décline en deux étapes principales. La première phase consiste en la construction d'une matrice de signatures mutuellement exclusive. Ainsi, le problème du nombre minimal de relations de redondance analytique (RRA), nécessaires pour établir un diagnostic sans ambiguïté, est abordé. Ce problème est formalisé comme un problème d'optimisation sous contraintes qui est efficacement résolu à l'aide d'un algorithme génétique. La deuxième étape concerne la génération des diagnostics. Ainsi, pour une situation observée, identifier les conflits revient à définir les RRAs non satisfaites par l'observation. Les diagnostics sont obtenus à l'aide d'un algorithme basé sur le concept de formules sous forme MNF (Maximal Normal Form). L'intérêt majeur dans cette approche est sa capacité à traiter le diagnostic des défauts simples et multiples ainsi que le diagnostic des plusieurs modes de fautes (i.e., le diagnostic des différents types de défauts) associés à chaque composant du système surveillé. De plus, il existe des preuves d'optimalité tant au niveau local (preuve de robustesse/sensibilité) qu'au niveau global (preuve de diagnostics minimaux). La méthodologie proposée est appliquée à la mission spatiale Mars Sample Return (MSR). Cette mission, entreprise conjointement entre l'administration nationale de l'aéronautique et de l'espace (NASA) et l'agence spatiale européenne (ESA), vise à ramener des échantillons martiens sur Terre pour des analyses. La phase critique de cette mission est la phase rendez-vous entre le conteneur d'échantillons et l'orbiteur. Les travaux de recherche traitent le problème de diagnostic des défauts capteurs présents sur la chaîne de mesure de l'orbiteur pendant la phase de rendez-vous de la mission. Les résultats, obtenus à l'aide du simulateur haute fidélité de Thalès Alenia Space, montrent la faisabilité et l'efficacité de la méthode. / The work presented in this thesis deals with the synthesis of algorithms for the diagnosis of simple and multiple faults. The main objective which is pursued is to design a fault diagnosis scheme by merging a minimum number of analytic redundancy with the available hardware redundancy. The main contribution of the proposed technique concerns the general architecture of the proposed diagnosis method. The originality of the research work is the combination of ideas and tools originated from two research communities : the FDI (Fault Detection and Isolation) community and the DX (Diagnosis) community whose foundations are derived from Computer Science and Artificial Intelligence fields. Hence, the fault detection problem (as well as the isolation task when structural constraints allow it) is solved by means of FDI techniques while the fault isolation problem is solved through the DX approaches, thus resulting in an aggregated methodology. The proposed method is divided in two steps. The first step deals with the construction of a mutually exclusive signature matrix. Hence, the problem of the minimal number of analytic redundancy relations (ARR), necessary for generating a diagnosis without any ambiguity, is treated. This problem is formalised as an optimized problem under constraints which is efficiently solved by means of a genetic algorithm. The second step concerns the generation of diagnoses. Thus, for an observed situation, the identification of conflicts results in the determination of the non satisfied ARRs for the given observation. The diagnoses are obtained by means an algorithm based on the concept of MNF (Maximal Normal Form) formulas. The main interest of this approach is its capacity to deal with the diagnosis of simple and multiple faults as well as the diagnosis of multi-modes faults (i.e., multiple types of faults) associated to each component of the system. Furthermore, it exists proofs on optimality both at a local level (proof of robustness/sensitivity) and at a global level (proof of minimal diagnoses). The proposed method is applied to the Mars Sample Return (MSR) mission. This spacecraft mission, undertaken jointly by the National Aeronautics and Space Administration (NASA) and the European Space Agency (ESA), aims at returning tangible samples from Mars atmosphere and ground to Earth for analysis. The critical phase of the mission is the rendezvous phase between the sample container vehicle and the orbiter spacecraft. The research work aims at realising sensor fault diagnosis on the orbiter during the rendezvous phase of the mission. Simulation results from the MSR high fidelity simulator, provided by Thalès Alenia Space, demonstrate the feasibility and the efficiency of the proposed approach.

La complexification des systèmes et la réduction du nombre de capteurs nécessitent l’élaboration de méthodes de surveillance de plus en plus efficaces. Le travail de cette thèse s’inscrit dans ce contexte et porte sur le diagnostic et le pronostic des Systèmes à Événements Discrets (SED) temporisés. Les réseaux de Petri stochastiques partiellement mesurés sont utilisés pour modéliser le système. Le modèle représente à la fois le comportement nominal et le comportement dysfonctionnel du système. Il permet aussi de représenter ses capteurs à travers une mesure partielle des transitions et des places. Notre contribution porte sur l’exploitation de l’information temporelle pour le diagnostic et le pronostic des SED. À partir d’une suite de mesures datées, les comportements du système qui expliqueraient ces mesures sont d’abord déterminés. La probabilité de ces comportements est ensuite évaluée pour fournir un diagnostic du système en termes de probabilité d’occurrence d’un défaut. Dans le cas où une faute est diagnostiquée, une approche permettant d’estimer la distribution de sa date d’occurrence est proposée. L’objectif est de donner plus de détails sur cette faute afin de mieux la caractériser. Par ailleurs, la probabilité des comportements compatibles est exploitée pour estimer l’état actuel du système. Il s’agit de déterminer les marquages compatibles avec les mesures ainsi que leurs probabilités associées. À partir de cette estimation d’état, la prise en considération des évolutions possibles du système permet d’envisager la prédiction de la faute avant son occurrence. Une estimation de la probabilité d’occurrence de la faute sur un horizon de temps futur est ainsi obtenue. Celle-ci est ensuite étendue à l’évaluation de la durée de vie résiduelle du système. Enfin, une application des différentes approches développées sur un cas d’un système de tri est proposée. / Due to the increasing complexity of systems and to the limitation of sensors number, developing monitoring methods is a main issue. This PhD thesis deals with the fault diagnosis and prognosis of timed Discrete Event Systems (DES). For that purpose, partially observed stochastic Petri nets are used to model the system. The model represents both the nominal and faulty behaviors of the system and characterizes the uncertainty on the occurrence of events as random variables with exponential distributions. It also considers partial measurements of both markings and events to represent the sensors of the system. Our main contribution is to exploit the timed information, namely the dates of the measurements for the fault diagnosis and prognosis of DES. From the proposed model and collected measurements, the behaviors of the system that are consistent with those measurements are obtained. Based on the event dates, our approach consists in evaluating the probabilities of the consistent behaviors. The probability of faults occurrences is obtained as a consequence. When a fault is detected, a method to estimate its occurrence date is proposed. From the probability of the consistent trajectories, a state estimation is deduced. The future possible behaviors of the system, from the current state, are considered in order to achieve fault prediction. This prognosis result is extended to estimate the remaining useful life as a time interval. Finally, a case study representing a sorting system is proposed to show the applicability of the developed methods.

Cette thèse s'intéresse à l'étude des problèmes de diagnostic des fautes sur les systèmes à événements discrets en utilisant les modèles réseau de Petri. Des techniques d'exploration incrémentale et à-la-volée sont développées pour combattre le problème de l'explosion de l'état lors de l'analyse de la diagnosticabilité. Dans le contexte atemporel, la diagnosticabilité de modèles RdP-L est abordée par l'analyse d'une série de problèmes K-diagnosticabilité. L'analyse de la diagnosticabilité est effectuée sur la base de deux modèles nommés respectivement FM-graph et FM-set tree qui sont développés à-la-volée. Un diagnostiqueur peut être dérivé à partir du FM-set tree pour le diagnostic en ligne. Dans le contexte temporel, les techniques de fractionnement des intervalles de temps sont élaborées pour développer représentation de l'espace d'état des RdP-LT pour laquelle des techniques d'analyse de la diagnosticabilité peuvent être utilisées. Sur cette base, les conditions nécessaires et suffisantes pour la diagnosticabilité de RdP-LT ont été déterminées. En pratique, l'analyse de la diagnosticabilité est effectuée sur la base de la construction à-la-volée d'une structure nommée ASG et qui contient des informations relatives à l'occurrence de fautes. D'une manière générale, l'analyse effectuée sur la base des techniques à-la-volée et incrémentale permet de construire et explorer seulement une partie de l'espace d'état, même lorsque le système est diagnosticable. Les résultats des simulations effectuées sur certains benchmarks montrent l'efficacité de ces techniques en termes de temps et de mémoire par rapport aux approches traditionnelles basées sur l'énumération des états / This PhD thesis deals with fault diagnosis of discrete event systems using Petri net models. Some on-the-fly and incremental techniques are developed to reduce the state explosion problem while analyzing diagnosability. In the untimed context, an algebraic representation for labeled Petri nets (LPNs) is developed for featuring system behavior. The diagnosability of LPN models is tackled by analyzing a series of K-diagnosability problems. Two models called respectively FM-graph and FM-set tree are developed and built on the fly to record the necessary information for diagnosability analysis. Finally, a diagnoser is derived from the FM-set tree for online diagnosis. In the timed context, time interval splitting techniques are developed in order to make it possible to generate a state representation of labeled time Petri net (LTPN) models, for which techniques from the untimed context can be used to analyze diagnosability. Based on this, necessary and sufficient conditions for the diagnosability of LTPN models are determined. Moreover, we provide the solution for the minimum delay ∆ that ensures diagnosability. From a practical point of view, diagnosability analysis is performed on the basis of on-the-fly building of a structure that we call ASG and which holds fault information about the LTPN states. Generally, using on-the-fly analysis and incremental technique makes it possible to build and investigate only a part of the state space, even in the case when the system is diagnosable. Simulation results obtained on some chosen benchmarks show the efficiency in terms of time and memory compared with the traditional approaches using state enumeration

Beim Entwurf großer Netzwerke der Automation ermöglichen Methoden der Leistungsbewertung den Test und die Dimensionierung eines Netzwerkes bereits vor der Inbetriebnahme und erlauben damit einen effizienten, qualitätsorientierten Entwurf. Es ist jedoch sehr wissensintensiv und aufwendig, entsprechende Analyse- oder Simulations-Modelle aufzustellen und die Ergebnisse auszuwerten, weshalb die Methoden in der Praxis der Automation selten verwendet werden. Viel vertrauter sind dem Entwerfer hingegen die speziellen Software-Tools, mit denen in der Automation Netzwerke entworfen werden. Auf Basis der Datenbanken dieser Tools werden in dieser Arbeit verschieden Verfahren zur automatischen Systemmodellierung, Verkehrsmodellierung, Leistungsbewertung und Fehlerdiagnose zu einem Control Network Performance Engineering kombiniert, welches die Qualitätsbewertung und -beratung nahtlos und ohne Mehraufwand in den Netzwerkentwurf integriert. (Die Dissertation wurde veröffentlicht im Jörg Vogt Verlag, Voglerstr. 20, 01277 Dresden, Internet: http://www.vogtverlag.de/, email: info@vogtverlag.de, ISBN 978-3-938860-10-6) / During the design of large automation networks, performance analysis methods can be used for testing and dimensioning the network before implementation and are essential for an efficient and reliable design process. However, setting up the necessary analytical or simulative models is time-consuming, requires in-depth knowledge, and is therefore often not applicable in practice. The network designers are much more used to the design tools used to develop automation networks. Based on these tools' databases various methods for automated system and traffic modeling, performance analysis and diagnoses are combined in the control network performance engineering that seamlessly integrates quality analysis and consulting into network design without requiring additional effort. (This manuscript is also available - in the form of a book - from Jörg Vogt Verlag, Voglerstr. 20, 01277 Dresden, Germany world-wide web address: http://www.vogtverlag.de/, electronic-mail address: info@vogtverlag.de, ISBN 978-3-938860-10-6 )

Wound Independently Controlled Stator Coil (WICSC) machines are a novel type of multiphase machine with the ability to change its pole-phase configuration during operation. This enables improved post-fault control and thus prolonged continuous operation. For a smooth transition to post-fault control, a quick and robust strategy for Fault detection and isolation (FDI) is essential. This master thesis concerns the design of a procedure to localise Open Phase Fault (OPF) and Open Switch Fault (OSF) in WICSC machines. The strategy is based on Harmonic Plane Decomposition (HPD), which generalises Vector Space Decomposition (VSD) to variable pole-phase machines. A previous iteration of the FDI strategy is analysed and developed further on MatLab. Experimental verification is performed on a WICSC motor with and without imitated OPF. A Printed Circuit Board (PCB) is designed to imitate OSF during operation. The outcome displays a fault detection within less than 30% of a fundamental current period. After no more than one period the full diagnosis is completed. Although only OPFs are tested, the results suggest the Fault detection and isolation (FDI)’s ability to identify OSFs as well. The parameter tuning is shown to be important for both correct fault localisation and avoiding false positives. Imbalances in the machine require special care when choosing the parameter values. Future work includes using the PCB for OSF tests and making the parameter tuning more robust. / WICSC-maskiner är en ny typ av multifasmaskin med förmågan att ändra sin polfas-konfiguration under drift. Detta möjliggör förbättrad reglering efter att fel uppstår och därmed förlängd drift. För att uppnå en mjuk övergång till reglering efter fel krävs en snabb och robust strategi för feldiagnosticering. Denna masteruppsats rör utvecklingen av en strategi för att detektera och lokalisera tomgångsfel i spolar och elektroniska brytare till WICSC-maskiner. Strategin baseras på HPD, som är en generalisering av VSD. En tidigare iteration av strategin analyseras och vidareutvecklas i MatLab. Experimentell verifiering utförs på en WICSC-motor med samt utan imiterade tomgångsfel i spolarna. En PCB designas för att imitera tomgångsfel i växelriktarens brytare. Resultaten ger en feldetektering på under 30% av en fundamental strömperiod. Efter maximalt en period är hela diagnosticeringen slutförd. Trots att endast spolfel testas pekar resultaten på att strategin även fungerar för brytarfel. Strategins parametervärden visar sig vara viktiga för både fellokaliseringen och för att undvika falska positiv. Obalans i maskinen ställer särsklit höga krav på justeringen av dessa värden. I framtida studier kommer PCB-enheten användas för att testa diagnosticering av brytarfel och göra justeringen av parametervärdena mer robust.

Tesis por compendio / [ES] Las averías inesperadas en las máquinas eléctricas rotativas pueden provocar tiempos de inactividad no programados, que pueden tener como consecuencia grandes pérdidas económicas. Esto aplica a las máquinas eléctricas rotativas en general y en particular a los aerogeneradores eólicos, cuya instalación y necesidades de mantenimiento han experimentado un gran crecimiento en los últimos años. Su mantenimiento predictivo evitaría las averías inesperadas y el aumento de costes que se pueden derivar de un mantenimiento preventivo, mediante la medida continua, on-line, de las variables que indican el estado del aerogenerador. Junto con ello, es necesario el desarrollo de técnicas de diagnóstico fiables que necesitan del testeo en aerogeneradores trabajando bajo diferentes condiciones de fallo. Como los ensayos destructivos para el estudio de defectos o fallos no es económicamente factible, se han desarrollado modelos de máquinas eléctricas rotativas trabajando en condiciones de fallo. Así, estos modelos permiten extraer las características de las corrientes de las máquinas bajo condiciones de fallo reduciendo el número de pruebas destructivas. A partir del compendio de artículos científicos y presentaciones en congreso de esta tesis doctoral se han desarrollado las diferentes etapas de implementación del banco de ensayos. En primer lugar, se exploran las diferentes técnicas de modelado de las máquinas eléctricas rotativas mediante una revisión exhaustiva de la literatura técnica disponible hasta el momento. Por otro lado, se pretenden desarrollar técnicas de modelado para ser implementadas en sistemas de prototipado rápido para realizar el diagnóstico online de la máquina. Sin embargo, los modelos analíticos como el presentado, debido a las simplificaciones que se asumen, no pueden modelar con precisión las no idealidades y las no idealidades propias de las máquinas de inducción defectuosas. Los modelos basados en métodos numéricos son más completos, pero requieren una gran capacidad computacional y largos tiempos de simulación. Además, su implementación en sistemas de prototipado rápido, resulta en una tarea muy compleja hasta el momento. Por esta razón, se opta por desarrollar un método de modelado basado en un enfoque híbrido analítico-numérico, que puede ayudar en el desarrollo de pruebas de técnicas de diagnóstico de fallos para ser implementados en dispositivos embebidos, así como para entrenar a sistemas expertos para evaluar la condición de la máquina. No obstante, este modelo desarrollado todavía requiere de importantes capacidades de memoria y tiempos de cómputo, por lo que se aporta un método para optimizar el cálculo de las inductancias de la máquina eléctrica. El estudio de esta técnica, además de una publicación en revista, ha resultado en dos presentaciones en congreso. Por otro lado, se ha optimizado el modelo analítico, aportando un nuevo método basado en el álgebra tensorial que además permite adaptar el modelo híbrido a diferentes tipos de defectos. Finalmente, como alternativa a las técnicas de diagnóstico tradicionales que se basan en el análisis de corriente a través de la transformada de Fourier (FFT), cuyo uso está limitado al diagnóstico en régimen estacionario, así como a las técnicas tiempo-frecuencia que permiten el diagnóstico de fallos bajo condiciones de régimen transitorio pero a un elevado coste computacional, este trabajo presenta una combinación de técnicas de diagnóstico que proporciona alta resolución espectral en todo el rango de carga de la máquina a bajo coste computacional y requisitos de memoria. Para la validación de las técnicas de modelado y la técnica de diagnóstico presentadas se ha implementado un banco de ensayos versátil con la capacidad de reproducir cualquier condición de funcionamiento. / [CA] Les averies inesperades de les màquines elèctriques rotatives poden provocar temps no programats de inactivitat, que esdevindrien en greus pèrdues econòmiques. Aquest raonament és cert per a màquines elèctriques rotatives en general i per als aerogeneradors en particular, la instal·lació dels quals i necessitats de manteniment han experimentat un gran creixement al llarg dels últims anys. El manteniment predictiu evitaria les averies inesperades i el augment de costos derivats de un manteniment preventiu, mitjançant el anàlisi continu i en línia de les magnituds que indiquen l'estat de l'aerogenerador. A més, seria necessari desenvolupar tècniques de diagnosi fiables a través de proves realitzades en aerogeneradors treballant amb distintes condicions de fallada. La realització d'assajos destructius per a l'estudi de defectes o averies no és econòmicament factible. Per tant, s'han desenvolupat models de màquines elèctriques rotatives treballant en condicions de fallada. D'aquesta manera, aquest model permeten extraure les característiques de les corrents absorbides/generades per les màquines en condicions de fallada i permeten reduir el nombre de proves destructives. A partir d'un compendi d'articles científics i presentacions en congrés d'aquesta tesi doctoral s'han desenvolupat les diferents etapes de implementació del banc de proves. En primer lloc s'exploren les diferents tècniques de modelat de màquines elèctriques rotatives mitjançant una revisió exhaustiva de la literatura tècnica disponible fins a la data. D'altra banda, es pretén desenvolupar tècniques de modelat per se implementades en sistemes de prototipat ràpid per realitzar la diagnosi en línia de la màquina. En aquest cas, la primera tècnica de modelat desenvolupada en la tesis es basa en mètodes analítics, obtinguin-se un model molt ràpid que simplifica el procés de càlcul dels paràmetres del model de la màquina elèctrica i és capaç de representar qualsevols tipus i nombre de fallades de simetria del rotor. No obstant això, els models analítics, como el presentat, no poden modelar amb precisió les no idealitats i no linealitats pròpies de les màquines de inducció amb fallada. El models basats en mètodes numèrics son més complets però requereixen una gran capacitat computacional i llargs temps de simulació. A més, la seua implantació en sistemes de prototipat ràpid resulta una tasca complexa i en molts casos inassolible. Per aquesta raó s'ha optat per desenvolupar un mètode de modelat basat en un plantejament híbrid analític-numèric, que pot ajudar al desenvolupament de sistemes per provar tècniques de diagnosi de fallades per se implantats en dispositius embeguts així com per entrenar a sistemes experts per avaluar la condició de la màquina. No obstant això, aquest model encara requereix de importants capacitats de memòria i temps de computació, per tant, en aquesta tesis, s'aporta un mètode per optimitzar el càlcul de les inductàncies de la màquina elèctrica. L'estudi d'aquesta tècnica, a més d'una publicació en revista, ha resultat en dos presentacions en congres. En altra publicació s'ha optimitzat el model analític mitjançant un mètode basat en l'àlgebra tensorial que permet adaptar el model híbrid a altres tipus de fallada. Finalment, com alternativa a les tècniques de diagnòstic tradicionals basades en el anàlisi de la corrent amb la transformada ràpida de Fourier (FFT) limitades al diagnòstic en règim estacionari, així como a les tècniques basades en l'anàlisi temps-freqüència que permeten la diagnosis en règim transitori per a un elevat cost computacional, aquest treball presenta una combinació de tècniques de diagnosi que proporciona alta resolució espectral en totes les zones de treball de la màquina amb un baix cost computacional i de memòria. Per a la validació d'aquestes tècniques de modelat i la tècnica de diagnosi presentada s'ha implementat un banc d'assajos versàtil i amb la capacitat de reproduir qualsevols condició de funcionament. / [EN] Unexpected breakdowns in rotating electrical machines components can lead to unscheduled downtimes and consequently to large economic losses. This relates to electrical machines in general and particularly to wind turbines, whose installation and maintenance needs have raised significantly in recent years. Predictive maintenance would avoid unexpected breakdowns and the associated increased costs from preventive maintenance, by online condition monitoring of the machine. In addition, the development of reliable diagnostic techniques is needed, which requires the testing of wind turbines working under various fault conditions. As destructive testing for fault research purposes is not economically feasible, several rotating electrical machines models running under fault conditions have been developed to investigate the characteristics of faulty machines and have allowed reducing the number of destructive tests. The different stages of the implementation of the test bench have been developed from the compendium of scientific papers and conference presentations of this doctoral thesis. Firstly, the different modeling techniques for rotating electrical machines are explored through an exhaustive review of the technical literature available so far. Moreover, it is aimed at developing modeling techniques valid to be implemented in rapid prototyping systems to perform online diagnosis of the machine. The first modeling technique developed in this thesis is based on analytical methods, obtaining a very fast model which greatly simplifies the process of calculating the parameters of the electrical induction machine model. This model can reproduce any kind and number of rotor asymmetry faults. However, analytical models as the presented one, due to the simplifications assumed, cannot accurately model the inherent non-idealities and non-idealities within faulty induction machines. Models based on numerical methods are more complete but require high computational effort and long simulation times. Moreover, their implementation in rapid prototyping systems is challenging so far. For this reason, a modeling method based on a hybrid analytical- numerical approach is developed, which can contribute to the development of testing fault diagnosis techniques to be implemented in embedded devices, as well as to train expert systems to evaluate the state of the machine. However, this model still requires a significant memory capacity and computation time, so it is provided a method to optimize the computation of the coupling para- meters of the machine. The study of this technique has resulted in a journal publication and two conference presentations. Particularly, it is studied the correct implementation of the modelling technique to obtain a reliable fault diagnosis and it is compared with another method of parameter reduction pro- posed in the technical literature. Besides that, the analytical model has been optimized, providing a new method based on tensor algebra, which also allows the hybrid model to be adapted to different types of defects. Finally, this work shows a combination of diagnostic techniques providing high spectral resolution over the entire machine load range at low computational cost and with negligible memory requirements. This provides an alternative to the traditional diagnostic techniques based on current analysis using Fourier transform (FFT), whose implementation is limited to the diagnosis in the steady state, as well as time-frequency techniques, which allow fault diagnosis under transient regime conditions at high computational cost. For the validation of the presented modeling techniques and diagnostic technique, a versatile test bench has been implemented. This test bench allows to reproduce any operating condition of the machine. / Terrón Santiago, C. (2022). Desarrollo e implementación de un banco de ensayos virtual de aerogeneradores para diferentes regímenes de funcionamiento y condiciones de fallo [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/183782 / TESIS / Compendio

Das Hauptmerkmal eines fehlertoleranten Regelungssystems ist die Aufrechterhaltung der Gesamtsystemstabilität und einer akzeptablen Leistung angesichts von Fehlern und Ausfällen innerhalb des Systems. In dieser Arbeit wird ein Verfahren zur Fehlererkennung und Ansätze der fehlertoleranten Regelung (FTR) mit einer Anwendung gegen Aktorfehler von Regelungssystemen, die aus sowohl einem einzelnen Aktor als auch mehreren Aktoren (z.B. Softrobotik) bestehen, vorgestellt. Diese Methode beruht hauptsächlich auf einem Index, genannt Fitnessindex (FI. Der FI wird durch den Vergleich der aktuellen Parameter des Aktors und die im Normalzustand mittels eines Modells geschätzt. Die Ergebnisse des Fitnessindex werden mit Hilfe des Performanceindex verwendet zur Bewertung des Schweregrades eines Aktorfehlers während des Betriebes in einem geschlossenen Regelkreis und sind die entscheidende Grundlage zur automatischen Fehlerdetektion, Fehlerdiagnose und der anschließenden FTR. Diese Methode wird in einem Simulationsmodell getestet, das dem Versuchsaufbau eines Formgedächtnislegierungs-Aktors entspricht. Die Simulationsergebnisse zeigen die Berechnungsergebnisse über den FI und anschließend die Verwendung der Ergebnisse des FIs zur Durchführung der FTR, um die Leistung zu gewährleisten und die Zuverlässigkeit der Regelungssysteme zu verbessern, wenn ein Fehler im Aktor auftritt. / The main feature of a fault tolerant control system is to maintain overall system stability and acceptable performance in the face of errors and failures within the system. In this thesis, a method for fault detection and approaches of the fault tolerant control (FTC) with an application against actuator errors of control systems, which consist of a single actuator as well as several actuators (soft robotics), is presented. This method is mainly based on an index called the fitness index (FI). The FI is estimated by comparing the current parameters of the actuator and in the normal state, based on a model. The results of the fitness index are used with the help of the performance index to evaluate the severity of an actuator fault during an operation in a control loop and are the decisive basis for automatic fault detection, fault diagnosis and subsequent the FTC. This method is tested in a simulation model that corresponds to the experimental setup of a shape memory alloy actuator. The simulation results show the calculation results about the FI and in addition the use the results of the FI to perform the FTC in order to ensure the performance and improve the reliability of the control systems when a fault occurs in the actuator.

Artificial Intelligence in Automotive Industry / Cost reduction is one of the main concerns in industry. Companies invest considerably for better performance in end-of-line fault diagnosis systems. A common strategy is to use data obtained from existing instrumentation. This research investigates the challenge of learning from historical data that have already been collected by companies. Machine learning is basically one of the most common and powerful techniques of artificial intelligence that can learn from data and identify fault features with no need for human interaction. In this research, labeled sound and vibration measurements are processed into fault signatures for vehicle starter motors and alternators. A fault detection and identification system has been developed to identify fault types for end-of-line testing of motors. However, labels are relatively difficult to obtain, expensive, time consuming and require experienced humans, while unlabeled samples needs less effort to collect. Thus, learning from unlabeled data together with the guidance of few labels would be a better solution. Furthermore, in this research, learning from unlabeled data with absolutely no human intervention is also implemented and discussed as well. / Thesis / Master of Applied Science (MASc)