Orientador: Fábio Roberto Chavarette / Resumo: Redes Neurais Artificiais (RNAs) são algoritmos de aprendizado, geralmente estruturados em torno de categorização de dados de entrada e/ou seu agrupamento por similaridade. Tendo em vista características desejáveis como aprendizado rápido e estabilidade frente a vetores de entrada altamente mutáveis, adotou-se uma RNA do tipo Fuzzy ARTMAP como mecanismo central de um método de monitoramento de saúde estrutural para detectar e categorizar falhas em dados experimentais provenientes de um sistema mecânico similar a um pequeno prédio de dois andares. Mais especificamente, com o objetivo de detectar alterações das frequências naturais da estrutura, fenômeno ligado à deterioração da mesma, e determinar qual(is) andar(es) está(ão) ligado(s) ao comportamento anômalo, se detectado. A acurácia da rede foi avaliada, sendo realizado um estudo da quantidade de dados necessárias para o desempenho satisfatório da rede. Observou-se desempenho satisfatório, a acurácia do método tendendo a aproximadamente 94% a partir de certas quantidades de dados. / Abstract: Artificial Neural Networks (ANNs) are learning algorithms, largely revolving around categorizing data sets based on measures of similarity between its members. Due to desirable characteristics such as fast learning and stability when dealing with highly mutable input vectors, a Fuzzy ARTMAP ANN was selected as the core mechanism of a structural health monitoring method. Its goal was to detect and categorize faults in experimental data collected from a mechanical system akin to a small two-story building. More specifically, to detect disturbances on the structure's natural frequencies, phenomenon linked to its deterioration, and to determine which story or stories are linked to anomalous behavior, if any. The accuracy of the method was evaluated, and the amount of data needed for optimal operation was determined. Satisfactory performance was observed; the method's accuracy tended towards 94% with enough training samples. / Mestre

[pt] Contratos de software podem ser escritos como expressões lógicas capazes de identificar falhas que ocorrem durante a utilização de um software. É possível implementar a verificação de um contrato em um software através de assertivas executáveis. No entanto, a forma como assertivas convencionais são implementadas não é diretamente aplicável a sistemas distribuídos, uma vez que apresentam dificuldades para avaliar expressões temporais, tampouco as expressões podem envolver propriedades de diferentes processos. Este trabalho propõe um mecanismo baseado em logs com meta-informações para a verificação de contratos em sistemas distribuídos. Uma gramática para redigir contratos possibilita operações temporais, ou seja, permite a especificação de condições entre eventos, em diferentes instantes de tempo, ou mesmo garante uma sequência de eventos, durante um período de tempo. O fluxo de eventos gerado é avaliado assincronamente em relação à utilização do sistema, pela comparação com contratos, previamente escritos de acordo com a gramática, que representam as expectativas sobre o comportamento normal do sistema. / [en] Software contracts can be written as assertions that identify failures observed while using the software. Software contracts can be implemented through executable assertions. However, conventional assertions are not directly applicable in distributed systems, as they present difficulties to evaluate temporal expressions, as well as expressions involving properties of different processes. This work proposes a mechanism based on logs with meta-information to evaluate contracts in distributed systems. A grammar to write contracts enable temporal operations, e.g., allows specifying conditions between events at different timestamps, or even guaranteeing a sequence of events over a period of time. The flow of events is evaluated asynchronously in relation to the system execution, by comparison with contracts, previously written according to the grammar, representing the expectations on the behavior of the system.

A presente tese apresenta proposições para o uso da seleção de variáveis no aprimoramento do controle estatístico de processos multivariados (MSPC) em bateladas, a fim de contribuir com a melhoria da qualidade de processos industriais. Dessa forma, os objetivos desta tese são: (i) identificar as limitações encontradas pelos métodos MSPC no monitoramento de processos industriais; (ii) entender como métodos de seleção de variáveis são integrados para promover a melhoria do monitoramento de processos de elevada dimensionalidade; (iii) discutir sobre métodos para alinhamento e sincronização de bateladas aplicados a processos com diferentes durações; (iv) definir o método de alinhamento e sincronização mais adequado para o tratamento de dados de bateladas, visando aprimorar a construção do modelo de monitoramento na Fase I do controle estatístico de processo; (v) propor a seleção de variáveis, com propósito de classificação, prévia à construção das cartas de controle multivariadas (CCM) baseadas na análise de componentes principais (PCA) para monitorar um processo em bateladas; e (vi) validar o desempenho de detecção de falhas da carta de controle multivariada proposta em comparação às cartas tradicionais e baseadas em PCA. O desempenho do método proposto foi avaliado mediante aplicação em um estudo de caso com dados reais de um processo industrial alimentício. Os resultados obtidos demonstraram que a realização de uma seleção de variáveis prévia à construção das CCM contribuiu para reduzir eficientemente o número de variáveis a serem analisadas e superar as limitações encontradas na detecção de falhas quando bancos de elevada dimensionalidade são monitorados. Conclui-se que, ao possibilitar que CCM, amplamente utilizadas no meio industrial, sejam adequadas para banco de dados reais de elevada dimensionalidade, o método proposto agrega inovação à área de monitoramento de processos em bateladas e contribui para a geração de produtos de elevado padrão de qualidade. / This dissertation presents propositions for the use of variable selection in the improvement of multivariate statistical process control (MSPC) of batch processes, in order to contribute to the enhacement of industrial processes’ quality. There are six objectives: (i) identify MSPC limitations in industrial processes monitoring; (ii) understand how methods of variable selection are used to improve high dimensional processes monitoring; (iii) discuss about methods for alignment and synchronization of batches with different durations; (iv) define the most adequate alignment and synchronization method for batch data treatment, aiming to improve Phase I of process monitoring; (v) propose variable selection for classification prior to establishing multivariate control charts (MCC) based on principal component analysis (PCA) to monitor a batch process; and (vi) validate fault detection performance of the proposed MCC in comparison with traditional PCA-based and charts. The performance of the proposed method was evaluated in a case study using real data from an industrial food process. Results showed that performing variable selection prior to establishing MCC contributed to efficiently reduce the number of variables and overcome limitations found in fault detection when high dimensional datasets are monitored. We conclude that by improving control charts widely used in industry to accomodate high dimensional datasets the proposed method adds innovation to the area of batch process monitoring and contributes to the generation of high quality standard products.

A diminuição no custo de computadores pessoais tem favorecido a construção de sistemas computacionais complexos, tais como aglomerados e grades. Devido ao grande número de recursos existentes nesses sistemas, a probabilidade de que faltas ocorram é alta. Uma abordagem que auxilia a tornar sistemas mais robustos na presença de faltas é a detecção de sua ocorrência, a fim de que processos possam ser reiniciados em estados seguros, ou paralisados em estados que não ofereçam riscos. Abordagens comumente adotadas para detecção seguem, basicamente, três tipos de estratégias: as baseadas em mensagens de controle, em estatística e em aprendizado de máquina. No entanto, elas tipicamente não consideram o comportamento de processos ao longo do tempo. Observando essa limitação nas pesquisas relacionadas, este trabalho apresenta uma abordagem para medir a variação no comportamento de processos ao longo do tempo, a fim de que mudanças inesperadas sejam detectadas. Essas mudanças são consideradas, no contexto deste trabalho, como faltas, as quais representam transições indesejadas entre estados de um processo e podem levá-lo a processamento incorreto, fora de sua especificação. A proposta baseia-se na estimação de cadeias de Markov que representam estados visitados por um processo durante sua execução. Variações nessas cadeias são utilizadas para identificar faltas. A abordagem proposta é comparada à técnica de aprendizado de máquina Support Vector Machines, bem como à técnica estatística Auto-Regressive Integrated Moving Average. Essas técnicas foram escolhidas para comparação por estarem entre as mais empregadas na literatura. Experimentos realizados mostraram que a abordagem proposta possui, com erro \'alfa\' = 1%, um F-Measure maior do que duas vezes o alcançado pelas outras técnicas. Realizou-se também um estudo adicional de predição de faltas. Nesse sentido, foi proposta uma técnica preditiva baseada na reconstrução do comportamento observado do sistema. A avaliação da técnica mostrou que ela pode aumentar em até uma ordem de magnitude a disponibilidade (em horas) de um sistema / The cost reduction for personal computers has enabled the construction of complex computational systems, such as clusters and grids. Because of the large number of resources available on those systems, the probability that faults may occur is high. An approach that helps to make systems more robust in the presence of faults is their detection, in order to restart or stop processes in safe states. Commonly adopted approaches for detection basically follow one of three strategies: the one based on control messages, on statistics or on machine learning. However, they typically do not consider the behavior of processes over time. Observing this limitation in related researches, this work presents an approach to measure the level of variation in the behavior of processes over time, so that unexpected changes are detected. These changes are considered, in the context of this work, as faults, which represent undesired transitions between process states and may cause incorrect processing, outside the specification. The approach is based on the estimation of Markov Chains that represent states visited by a process during its execution. Variations in these chains are used to identify faults. The approach is compared to the machine learning technique Support Vector Machines, as well as to the statistical technique Auto-Regressive Integrated Moving Average. These techniques have been selected for comparison because they are among the ones most employed in the literature. Experiments conducted have shown that the proposed approach has, with error \'alpha\'= 1%, an F-Measure higher than twice the one achieved by the other techniques. A complementary study has also been conducted about fault prediction. In this sense, a predictive approach based on the reconstruction of system behavior was proposed. The evaluation of the technique showed that it can provide up to an order of magnitude greater availability of a system in terms of uptime hours

Identity based cryptography was first proposed by Shamir in 1984. Rather than deriving a public key from private information, which would be the case in traditional public key encryption schemes, in identity based schemes a user's identity plays the role of the public key. This reduces the amount of computations required for authentication, and simplifies key-management. Efficient and strong implementations of identity based schemes are based around easily computable bilinear mappings of two points on an elliptic curve onto a multiplicative subgroup of a field, also called pairing. The idea of utilizing the identity of the user simplifies the public key infrastructure. However, since pairing computations are expensive for both area and timing, the proposed identity based cryptosystem are hard to implement. In order to be able to efficiently utilize the idea of identity based cryptography, there is a strong need for an efficient pairing implementations. Pairing computations could be realized in multiple fields. Since the main building block and the bottleneck of the algorithm is multiplication, we focused our research on building a fast and small arithmetic core that can work on multiple fields. This would allow a single piece of hardware to realize a wide spectrum of cryptographic algorithms, including pairings, with minimal amount of software coding. We present a novel unified core design which is extended to realize Montgomery multiplication in the fields GF(2^n), GF(3^m), and GF(p). Our unified design supports RSA and elliptic curve schemes, as well as identity based encryption which requires a pairing computation on an elliptic curve. The architecture is pipelined and is highly scalable. The unified core utilizes the redundant signed digit representation to reduce the critical path delay. While the carry-save representation used in classical unified architectures is only good for addition and multiplication operations, the redundant signed digit representation also facilitates efficient computation of comparison and subtraction operations besides addition and multiplication. Thus, there is no need for transformation between the redundant and non-redundant representations of field elements, which would be required in classical unified architectures to realize the subtraction and comparison operations. We also quantify the benefits of unified architectures in terms of area and critical path delay. We provide detailed implementation results. The metric shows that the new unified architecture provides an improvement over a hypothetical non-unified architecture of at least 24.88 % while the improvement over a classical unified architecture is at least 32.07 %. Until recently there has been no work covering the security of pairing based cryptographic hardware in the presence of side-channel attacks, despite their apparent suitability for identity-aware personal security devices, such as smart cards. We present a novel non-linear error coding framework which incorporates strong adversarial fault detection capabilities into identity based encryption schemes built using Tate pairing computations. The presented algorithms provide quantifiable resilience in a well defined strong attacker model. Given the emergence of fault attacks as a serious threat to pairing based cryptography, the proposed technique solves a key problem when incorporated into software and hardware implementations. In this dissertation, we also present an efficient accelerator for computing the Tate Pairing in characteristic 3, based on the Modified Duursma Lee algorithm.

Because of the increasingly demanding tasks that robotic systems are asked to perform, there is a need to make them more reliable, intelligent, versatile and self-sufficient. Furthermore, throughout the robotic system?s operation, changes in its internal and external environments arise, which can distort trajectory tracking, slow down its performance, decrease its capabilities, and even bring it to a total halt. Changes in robotic systems are inevitable. They have diverse characteristics, magnitudes and origins, from the all-familiar viscous friction to Coulomb/Sticktion friction, and from structural vibrations to air/underwater environmental change. This thesis presents an on-line environmental Change, Detection, Isolation and Accommodation (CDIA) scheme that provides a robotic system the capabilities to achieve demanding requirements and manage the ever-emerging changes. The CDIA scheme is structured around a priori known dynamic models of the robotic system and the changes (faults). In this approach, the system monitors its internal and external environments, detects any changes, identifies and learns them, and makes necessary corrections into its behavior in order to minimize or counteract their effects. A comprehensive study is presented that deals with every stage, aspect, and variation of the CDIA process. One of the novelties of the proposed approach is that the profile of the change may be either time or state-dependent. The contribution of the CDIA scheme is twofold as it provides robustness with respect to unmodeled dynamics and with respect to torque-dependent, state-dependent, structural and external environment changes. The effectiveness of the proposed approach is verified by the development of the CDIA scheme for a SCARA robot. Results of this extensive numerical study are included to verify the applicability of the proposed scheme.

Parmi les méthodes de diagnostic de défauts issue de l'automatique, la méthode des filtres résiduels permet de synthétiser des filtres générant des signaux (dits résidus) qui sont utilisés à des ns de détection de défauts. Dans notre approche, les filtres résiduels sont obtenus à partir d'observateurs réduits. L'objectif decette thèse est de synthétiser des observateurs et de mettre en évidence leur application à la détection de défauts pour les systèmes non linéaires. Ce mémoire est réparti en deux parties. Dans la première partie, deux contributions ont été présentées. La première concerne les observateurs à entrées inconnues pour les systèmes affines en l'état modulo une injection de sortie. L'approche proposée est une combinaison des techniques de découplage géométrique et des observateurs non linéaires. Des conditions suffisantes ont été données accompagnées d'un algorithme permettant de concevoir un observateur à entrées inconnues permettant d'estimer une partie de l'état indépendamment de la connaissance de certaines entrées. La deuxième contribution consiste à caractériser la classe des systèmes non linéaires qui se transforment en des systèmes en cascade pour lesquels un observateur peut être conçu. Tout d'abord, des conditions nécessaires et suffisantes théoriques ont été données, ensuite un algorithme permettant de calculer ces transformations (diéomorphismes) a été proposé. Enfin, l'ensemble de tous ces difféomorphismes a été caractérisé en montrant que ce dernier est une orbite d'une action d'un groupe particulier sur l'ensemble de tous les difféomorphismes. La deuxième partie de cette thèse concerne la synthèse d'un filtre polytopique Linéaires à Paramètres Variants (LPV) permettant de détecter, isoler et estimer de multiples défauts capteur. L'avantage de ce filtre est de générer deux types de résidus : l'un étant sensible aux défauts et l'autre insensible. Le résidu insensible est utilisé pour fournir une information qualitative supplémentaire de l'efficacité du filtre. La stabilité de ce dernier est analysée au moyen d'Inégalités Matricielles Linéaires (LMI) / Among the faults diagnosis methods, the method of residual filters allows to synthesize filters generating signals ( called residues) that are used for fault detection. In our approach, the residual filters are obtained from reduced observers. The objective of this thesis is to synthesize observers and highlight their application to fault detection for nonlinear systems. This thesis is divided into two parts. In the first part, two papers were presented. The rst one relates to the unknown input observers for state ane systems modulo an output injection. The proposed approach is a combination of geometric decoupling techniques and nonlinear observers. We have given sufficient conditions with an algorithm for designing an unknown input observer to estimate a part of the state without the knowledge of some inputs. The second contribution consists to characterize the class of nonlinear systems which can be transformed into cascade systems for which an observer can be designed. First, necessary and sufficient theoretical conditions were given, then an algorithm to compute these transformations (diffeomorphisms) was proposed. Finally, the set of all dieomorphisms was characterized by showing that it is an orbit of an action of a particular group on the set of all dieomorphisms. The second step of the thesis deals with the synthesis of a polytopic Linear Parameter-Varying (LPV) filter to detect, isolate and estimate multiple sensor faults. The advantage of this lter is to generate two types of residuals : one being sensitive to faults and the other is insensitive. The insensitive residual is used to generate an additional qualitative information of the filter efficiency. The stability of the latter has been performed using Linear Matrix Inequality (LMI)

As the world moves towards a cleaner and greener future, electrical machines for various industrial purposes and transport applications have gained a lot of attention. Permanent magnet synchronous machines (PMSMs) are usually the solution for electric vehicle (EV) applications thanks to their high efficiency, compactness and high-power density. On the downside, although the price of rare-earth materials has recovered close to historical levels, concerns still remain and the questions on the environmental sustainability of these materials have also been raised, which has encouraged the researchers to consider rare-earth-free machines.The switched reluctance machine (SRM) is one of the competitive alternatives, thanks to the simple and robust construction, high reliability and inherent fault tolerance capability. However, it has a bad reputation when it comes to torque ripple and acoustic noise. And the highly nonlinear characteristic brings much difficulty to routine design purposes and machine optimisation.Therefore, some of the above mentioned problems are addressed - a torque-ripple-reduction, reliable and low-cost system of SRMs is presented in this thesis. Firstly from the modelling point of view, a combined magnetic equivalent circuit (MEC) and finite element (FE) model of SRMs is developed for fast characterization the nonlinear behavior. Secondly from the control point of view, various torque-ripple reduction techniques are implemented and compared. Moreover, a minimal current sensing strategy with enhanced fault-detection capability is proposed and validated experimentally. It requires two current sensors, to replace the phase current sensors, with no additional devices for fault detection, to achieve a more compact and low-cost drive. Finally from the reliability point of view, an interturn short-circuit fault detection method and a rotor position estimation approach are investigated and validated experimentally, which leads to a more reliable system. / Doctorat en Sciences de l'ingénieur et technologie / info:eu-repo/semantics/nonPublished

Existem inúmeras ferramentas e metodologias para o diagnóstico de falhas em transformadores de potência, tanto para a monitoração e acompanhamento do equipamento em operação (planta fixa) como àquele retirado e abrigado em laboratórios de unidades fabris. Em função dos custos envolvidos no transporte dos equipamentos, manuseio para a substituição, energia não faturada nos períodos de manobras e custos adicionais dos ensaios para voltar com o equipamento em operação, as principais metodologias que se destacam são aquelas direcionadas ao equipamento em operação e também as escolhidas para serem analisadas neste trabalho. Após um estudo sistemático dessas metodologias, tais como termográfica, emissão acústica e análise dos gases dissolvidos no óleo, denominada cromatografia, verificou-se que esta última se destaca como a mais econômica e a mais difundida na identificação das falhas. Contudo, na cromatografia, existem diferentes critérios de análise baseados nas relações e quantidades de diferentes tipos de gases e que são usados pelas companhias concessionárias de forma indiscriminada, gerando muitas dúvidas quanto à sua validade. Esta pesquisa teve como principal foco a comparação desses critérios usando como base de dados àqueles fornecidos pela IEC e pela Companhia Paulista de Força e Luz (CPFL). A base de dados da CPFL contem aproximadamente quatro mil ensaios cromatográficos, colhidas nas ultimas três décadas correspondendo a 500 unidades transformadoras, nas potências de 5,0 a 50 MVA, instaladas em subestações nas tensões primárias de 69 e 138 kV, e secundária de 13,8 kV. Também fez parte dessa pesquisa a definição de um conjunto de critérios que melhor identificam as falhas em transformadores. / There are numerous tools and methodologies for fault diagnosis in power transformers, either for monitoring and tracking equipment in operation (fixed plant) or for those removed and housed in laboratories plants. Due to the costs involved in transporting the equipment, handling for the replacement, unbilled energy during periods of maneuvers and additional costs of the tests to return with the equipment in operation, the main methodologies that stand out are those using the equipment in operation and also those chosen to be analyzed in this work. After a systematic study of these methods such as thermography, and acoustic emission and analysis of gases dissolved in the oil, known as chromatography, it was found that the latter stands out as the most economical and most widely in the identification of faults. However, in chromatography, there are different criteria based on the relationships of different types and quantities of gases that are used by electrical companies indiscriminately, raising many questions about its validity. This research was mainly focused on the comparison of these criteria using the database as those provided by IEC and the Companhia Paulista de Força e Luz (CPFL). The database of CPFL contains approximately four thousand chromatographic assays from the last three decades, corresponding 500 transforming units, from 5 to 50 MVA, in substations with primary voltages of 138 kV and 69 and secondary of 13,8 kV. Was also studied the definition of a set of criteria which identify faults in transformers.

O objetivo deste trabalho é estudar a identificação, classificação, localização e setorização de faltas em redes de distribuição radiais, verificar a maneira de aplicar e integrar diversas ferramentas numéricas convencionais, assim como ferramentas de sistemas inteligentes, visando identificar a ocorrência de uma falta, classificar as fases envolvidas com a mesma, e aplicar as diversas ferramentas existentes a fim de localizar em tempo real o eventual local onde houve a ocorrência da falta, permitindo que a mesma possa ser setorizada dentro do sistema da concessionária e informada ao centro de operações, usando uma nova abordagem inteligente baseada em análise espectral de sinais. / The aim of this work is study the identification, classification, location and sectorization of a fault in distribution radial networks, check how to implement and integrate various conventional numerical tools, as well as intelligent systems based tools, to identify the occurrence of a fault, classify the phases involved with it, and apply the various tools available to locate the place where a fault was occurred in real time, enabling it to be sectorized into the utility system and informed to operational center using a new intelligent approach based on spectral signals analysis.