Nesta dissertação é apresentada a comparação do desempenho de três estimadores - o Filtro de Kalman Estendido, o Filtro de Kalman Unscented e o Filtro de Partículas - aplicados para estimar a postura de um robô diferencial. Uma câmera foi fixa no teto para cobrir todo o campo operacional do robô durante os experimentos, a fim de extrair o mapa e gerar o ground truth. Isso permitiu realizar uma análise do erro de forma precisa a cada instante de tempo. O desempenho de cada um dos estimadores foi avaliado sistematicamente e numericamente para duas trajetórias. Os resultados desse primeiro experimento demonstram que os filtros proporcionam grandes melhorias em relação à odometria e que o modelo dos sensores é crítico para obter esse desempenho. O Filtro de Partículas mostrou um desempenho melhor em relação aos demais nos dois percursos. No entanto, seu elevado custo computacional dificulta sua implementação em uma aplicação de tempo real. O Filtro de Kalman Unscented, por sua vez, mostrou um desempenho semelhante ao Filtro de Kalman Estendido durante a primeira trajetória. Porém, na segunda trajetória, a qual possui uma quantidade maior de curvas, o Filtro de Kalman Unscented mostrou uma melhora significativa em relação ao Filtro de Kalman Estendido. Foi realizado um segundo experimento, em que o robô planeja e executa duas trajetórias. Os resultados obtidos mostraram que o robô consegue chegar a um determinado local com uma precisão da mesma ordem de grandeza do que a obtida durante a estimação de estados do robô. / In this dissertation, the performance of three nonlinear-model based estimators - the Extended Kalman Filter, the Unscented Kalman Filter and the Particle Filter - applied to pose estimation of a differential drive robot is compared. A camera was placed above the operating field of the robot to record the experiments in order to extract the map and generate the ground truth so the evaluation of the error can be done at each time step with high accuracy. The performance of each estimator is assessed systematically and numerically for two robot trajectories. The first experimental results showed that all estimators provide large improvements with respect to odometry and that the sensor modeling is critical for their performance. The particle filter showed a better performance than the others on both experiments, however, its high computational cost makes it difficult to implement in a real-time application. The Unscented Kalman Filter showed a similar performance to the Extended Kalman Filter during the first trajectory. However, during the second one (a curvier path) the Unscented Kalman Filter showed a significant improvement over the Extended Kalman Filter. A second experiment was carried out where the robot plans and executes a trajectory. The results showed the robot can reach a predefined location with an accuracy of the same order of magnitude as the obtained during the robot pose estimation.

abstract: In modern electric power systems, energy management systems (EMSs) are responsi-ble for monitoring and controlling the generation system and transmission networks. State estimation (SE) is a critical `must run successful' component within the EMS software. This is dictated by the high reliability requirements and need to represent the closest real time model for market operations and other critical analysis functions in the EMS. Tradi-tionally, SE is run with data obtained only from supervisory control and data acquisition (SCADA) devices and systems. However, more emphasis on improving the performance of SE drives the inclusion of phasor measurement units (PMUs) into SE input data. PMU measurements are claimed to be more accurate than conventional measurements and PMUs `time stamp' measurements accurately. These widely distributed devices meas-ure the voltage phasors directly. That is, phase information for measured voltages and currents are available. PMUs provide data time stamps to synchronize measurements. Con-sidering the relatively small number of PMUs installed in contemporary power systems in North America, performing SE with only phasor measurements is not feasible. Thus a hy-brid SE, including both SCADA and PMU measurements, is the reality for contemporary power system SE. The hybrid approach is the focus of a number of research papers. There are many practical challenges in incorporating PMUs into SE input data. The higher reporting rates of PMUs as compared with SCADA measurements is one of the salient problems. The disparity of reporting rates raises a question whether buffering the phasor measurements helps to give better estimates of the states. The research presented in this thesis addresses the design of data buffers for PMU data as used in SE applications in electric power systems. The system theoretic analysis is illustrated using an operating electric power system in the southwest part of the USA. Var-ious instances of state estimation data have been used for analysis purposes. The details of the research, results obtained and conclusions drawn are presented in this document. / Dissertation/Thesis / M.S. Electrical Engineering 2013

High precision control of a truck and trailer system requires accurate and robust state estimation of the system. This thesis work explores the possibility of estimating the states with high accuracy from sensors solely mounted on the truck. The sensors used are a LIDAR sensor, a rear-view camera and a RTK-GNSS receiver. Information about the angles between the truck and the trailer are extracted from LIDAR scans and camera images through deep learning and through model-based approaches. The estimates are fused together with a model of the dynamics of the system in an Extended Kalman Filter to obtain high precision state estimates. Training data for the deep learning approaches and data to evaluate and compare these methods with the model-based approaches are collected in a simulation environment established in Gazebo. The deep learning approaches are shown to give decent angle estimations but the model-based approaches are shown to result in more robust and accurate estimates. The flexibility of the deep learning approach to learn any model given sufficient training data has been highlighted and it is shown that a deep learning approach can be viable if the trailer has an irregular shape and a large amount of data is available. It is also shown that biases in measured lengths of the system can be remedied by estimating the biases online in the filter and this improves the state estimates.

O processo de estimação de estado em sistemas elétricos de potência está sujeito a três tipos de erros: erros nas medidas analógicas (erros grosseiros); erros devido a informações erradas quanto aos estados de chaves e/ou disjuntores (erros topológicos) e erros causados por informações erradas de algum parâmetro do sistema (erros de parâmetros). É drástico o efeito de erros de parâmetros, para o processo de estimação de estado, normalmente intolerável, sendo, entretanto, menos evidente que os erros grosseiros e topológicos. Aproveitando o fato de que certas medidas não sofrem mudanças significativas de valor, durante um determinado intervalo de tempo, propõe-se uma metodologia para estimação de estado e parâmetros de linhas de transmissão. Na metodologia proposta, que se baseia nas equações normais, o vetor de estado convencional é aumentado para a inclusão dos parâmetros a serem estimados. Este vetor de estado aumentado é então estimado através de uma grande quantidade de medidas, obtidas em diversas amostras, durante um intervalo de tempo em que as variáveis de estado do sistema não tenham sofrido alterações significativas de valor. Esta situação ocorre tipicamente à noite, fora dos horários de pico. Propõe-se também uma metodologia para análise de observabilidade para o estimador proposto. Para comprovar a eficiência das metodologias propostas, vários testes foram realizados, utilizando os sistemas de 6, 14 e 30 barras do IEEE. / The process of power system state estimation is subjected to three types of errors: errors in analogical measurements (gross errors), incorrect information about the status of switching devices (topology errors) and incorrect information about the model of the systems equipment (parameter errors). The effects of parameter errors on the process of power system state estimation are drastic and less evident to detect than gross and topology errors. Taking advantage of the fact that a certain fraction of the measurements varies over a small range in a certain period of time, a methodology to estimative transmission line parameters and state based on normal equations has been proposed. In such methodology, which is based on normal equations, the traditional state vector is expanded to include the parameters to be estimated. This augmented state vector is estimated through a large collection of measurements, recorded within several snapshots of the power system, during which the actual system state varies over a small range. This situation typically occurs during the night off-peak periods. An observability analysis methodology is also proposed for the presented estimator. To prove the efficiency of the methodologies, several tests were made using the systems of 6, 14 and 30 buses from IEEE.

Orientador: Madson Cortes de Almeida / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-24T01:13:38Z (GMT). No. of bitstreams: 1 Dardengo_VictorPellanda_M.pdf: 2575363 bytes, checksum: 30d6d4f10eba96b3baa38c14dae17316 (MD5) Previous issue date: 2013 / Resumo: No planejamento de sistemas de medição para estimação de estado de sistemas de energia elétrica são definidos os tipos, a localização e a quantidade de medidores. Um plano de medição adequado deve garantir a observabilidade da rede, viabilizando a estimação do estado de toda a rede, e permitir a filtragem de erros grosseiros. Para que os erros grosseiros possam ser filtrados é necessário que no plano de medição não estejam presentes medidas e conjuntos críticos. Neste contexto, esta dissertação propõe uma nova metodologia de análise de observabilidade baseada no estimador linearizado regularizado e apresentam duas técnicas de classificação de medidas capazes de identificar medidas críticas, medidas pertencentes a conjuntos críticos e medidas redundantes. A primeira técnica é baseada na fatoração da matriz Gram das medidas e a segunda é baseada na fatoração da matriz Jacobiana das medidas. Além disso, são estudadas e aplicadas técnicas de fatoração com números inteiros que conferem maior robustez 'as metodologias de classificação de medidas. São apresentados testes realizados em redes de pequeno e médio porte que apontam o bom funcionamento dos métodos desenvolvidos e estudados / Abstract: In planning of power systems state estimation are defined the type, the location and the number of meters. An adequate measurement plan have to ensure the observability of the network, enabling the state estimation of the entire network, and allow filtering of gross errors. In order to filter gross errors is necessary in the measurement plan is not present critical measurements and critical sets. In this context, this work proposes a new observability analysis methodology based on the regularized weighted least squares DC state estimator and presents two measurements classification techniques capable of identifying critical measurements, measurements belonging to critical sets and redundant measurements. The first technique is based on the factorization of the measurement Gram matrix and second is based on the factorization of the measurement Jacobian matrix. Furthermore, are studied and applied factorization techniques with integers numbers which give greater robustness to measurements classification methodologies. Tests are presented in small and medium networks size showing the proper functioning of developed and studied methods / Mestrado / Energia Eletrica / Mestre em Engenharia Elétrica

This work takes place in the context of distribution grids with high level of distributed generation, for example in microgrids. With high level of distributed generation, it has been shown that selective, fast and sensitive network protection is expected to be more difficult. Furthermore, during system restoration, the accurate fault location could be more challenging to assess, thereby increasing the average outage duration.<p>Thanks to cost reductions and improvement of information and communication technologies, future distribution networks will probably have advanced communication infrastructures and more measurement devices installed in order to manage the increasing complexity of those networks, which is primarily caused by the introduction of distributed generation at the distribution level.<p>Therefore this thesis investigates how the monitoring, protection and fault location functions can be improved by using system-wide measurements, i.e. real-time measurements such as synchronized voltage and current measurements recorded at different network locations. Distributed synchronized measurements bring new perspectives for these three functions: protection and fault location are usually performed with local measurements only and synchronized measurements are not common in monitoring applications. For instance, by measuring distributed generators infeed together with some feeder measurements, the protection is expected to be more sensitive and selective and the fault location to be more accurate.<p>The main contribution of this work is the use of state estimation, which is normally only used for network monitoring, for the protection and the fault location. <p>The distribution system state estimation is first developed using the classical transmission system approach. The impact of the placement of the measurement devices and of a relatively low measurement redundancy on the accuracy, on the bad data detection and on the topology error identification capabilities of the estimator are discussed and illustrated. This results in recommendations on the placement of the meters.<p>Then, a backup protection algorithm using system-wide measurements is presented. The coherence of the measurements and the healthy network model are checked thanks to a linear three-phase state estimation. If the model does not fit to the measurements and if the estimated load is too high or unbalanced, a fault is detected. The advantages of the method are that the voltage measurement redundancy is considered, improving the detection sensitivity, and that load models may be considered in the algorithm, avoiding the need to install measurement devices on every line of the network.<p>Finally, two new impedance-based fault location algorithms using distributed voltage and current recordings are proposed. By defining statistical errors on the measurements and the network parameters, a method to compute a confidence interval of the fault distance estimate is proposed. The fault location accuracy and its sensitivity to the fault conditions (e.g. fault resistance or fault type) and to the different sources of error are assessed on a simulated distribution system. / Doctorat en Sciences de l'ingénieur / info:eu-repo/semantics/nonPublished

State estimation is an important tool for power system monitoring and the present study involves integrating phasor measurement units in the state estimation process. Based on measurements taken throughout the network, the role of a state estimator is to estimate the state variables of the power system while checking that these estimates are consistent with the measurement set. In the case of power system state estimation, the state variables are the voltage phasors at each network bus.\\<p><p>The classical state estimator currently used is based on SCADA (Supervisory Control and Data Acquisition) measurements. Weaknesses of the SCADA measurement system are the asynchronicity of the measurements, which introduce errors in the state estimation results during dynamic events on the electrical network.\\<p><p>Wide-area monitoring systems, consisting of a network of Phasor Measurement Units (PMU) provide synchronized phasor measurements, which give an accurate snapshot of the monitored part of the network at a given time. The objective of this thesis is to integrate PMU measurements in the state estimator. The proposed state estimators use PMU measurements exclusively, or both classical and PMU measurements.\\ <p><p>State estimation is particularly useful to filter out measurement noise, detect and eliminate bad data. A sensitivity analysis to measurement errors is carried out for a state estimator using only PMU measurements and a classical state estimator. Measurement errors considered are Gaussian noise, systematic errors and asynchronicity errors. Constraints such as zero injection buses are also integrated in the state estimator. Bad data detection and elimination can be done before the state estimation, as in pre-estimation methods, or after, as in post-estimation methods. For pre-estimation methods, consistency tests are used. Another proposed method is validation of classical measurements by PMU measurements. Post-estimation is applied to a measurement set which has asynchronicity errors. Detection of a systematic error on one measurement in the presence of Gaussian noise is also analysed. \\<p><p>The state estimation problem can only be solved if the measurements are well distributed over the network and make the network observable. Observability is crucial when trying to solve the state estimation problem. A PMU placement method based on metaheuristics is proposed and compared to an integer programming method. The PMU placement depends on the chosen objective. A given PMU placement can provide full observability or redundancy. The PMU configuration can also take into account the zero injection nodes which further reduce the number of PMUs needed to observe the network. Finally, a method is proposed to determine the order of the PMU placement to gradually extend the observable island. \\<p><p>State estimation errors can be caused by erroneous line parameter or bad calibration of the measurement transformers. The problem in both cases is to filter out the measurement noise when estimating the line parameters or calibration coefficients and state variables. The proposed method uses many measurement samples which are all integrated in an augmented state estimator which estimates the voltage phasors and the additional parameters or calibration coefficients. / Doctorat en Sciences de l'ingénieur / info:eu-repo/semantics/nonPublished

Orientador: Madson Cortes de Almeida / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-27T18:54:32Z (GMT). No. of bitstreams: 1 YucraCcahuana_MiguelAngel_M.pdf: 1685093 bytes, checksum: d470831fabdff66aa7a9d998da9391db (MD5) Previous issue date: 2015 / Resumo: Nas últimas décadas, a utilização de Unidades de Medições Fasoriais (Phasor Measurements Units - PMU) em sistemas de transmissão de energia elétrica vem sendo amplamente discutida. De acordo com a literatura, diversas aplicações podem ser beneficiadas pelo uso destas medidas. Em comparação com as medidas convencionais disponibilizadas pelos Sistema de Controle e Aquisição de Dados (SCADA), as medições fasoriais apresentam maior precisão, maior taxa de amostragem e são grandezas fasoriais sincronizadas, enquanto que as medições do sistema SCADA são medidas de módulo, não sincronizadas e de menor precisão. Para que as aplicações em regime permanente possam se beneficiar das medidas providas pelas PMUs, essas medidas devem ser filtradas por um processo de estimação de estado. Apesar da precisão e da confiabilidade, as medidas provenientes de PMUs ainda podem conter erros causados no processo de medição, os quais precisam ser eliminados antes que as informações providas pelas PMUs sejam usadas nas tarefas de controle e operação da rede. Apesar de todos os investimentos e desenvolvimentos relacionados às PMUs, atualmente, devido ao grande custo associado à instalação dessas unidades, não se considera viável o desenvolvimento de sistemas de medição que permitam que redes sejam totalmente observáveis apenas com o uso de medições fasoriais. A literatura apresenta algumas técnicas capazes de considerar simultaneamente medidas fasoriais e medidas convencionais no processo de estimação de estado. Neste trabalho, são apresentados e avaliados os principais estimadores de estado de uma e duas fases disponíveis na literatura. Além disso, o estimador tradicional, sem a presença de medidas fasoriais, foi usado como base de comparação. Na primeira etapa do processo de avaliação é analisada a qualidade dos estimadores de estado na presença de medidas perfeitas e na presença de medidas contendo erros gaussianos aleatórios. Erros gaussianos aleatórios são os erros típicos dos processos de medição e, portanto, são considerados aceitáveis. Na segunda etapa do processo de avaliação, os estimadores são avaliados na presença de medidas contendo erros gaussianos aleatórios e erros grosseiros. Os erros grosseiros, ao contrário dos erros gaussianos, não são considerados aceitáveis e devem ser removidos durante o processo de estimação de estado. A fim de tornar a análise mais ampla e realista, os três principais mecanismos de cálculo de desvios padrão propostos na literatura são considerados na avaliação do estimadores. Assim, são considerados cenários com desvios padrão fixos, desvios padrão calculados em função dos valores das medidas e desvios padrão calculados em função dos valores medidos e dos fundos de escala dos medidores. Para a detecção e identificação de erros grosseiros é adotado o método do Maior Resíduo Normalizado, já que este é o mecanismo mais confiável, robusto e aceito para tal finalidade. São apresentados testes realizados na rede de 14 barras do IEEE. Esta rede é amplamente adotada em teses e artigos especializados da área, o que facilita a comparações de resultados / Abstract: In last decades, the use of Phasor Measurement Units (PMU) in electrical power transmission systems has been widely discussed. According to the literature, several applications may be benefited with the use of these measurements. When they are compared with conventional measurements provided by Supervisory control and data adquisition (SCADA), phasor measurements are more accurate, have a high sampling rate and are synchronized, while the SCADA measurements only provide magnitude measurements, are not synchronized and are less accurate. For applications in steady operation can be benefited from the measurements provided by PMUs, these measurements must be filtered by a state estimation process. Despite the accuracy and reliability, measurements from PMUs may contain errors caused by the measurement process, they must be eliminated before the informations provided by PMUs are used in control tasks and network operation. Moreover, despite all the investments and developments related to PMUs, currently, because of the cost associated to the installation of these units; it is not considered feasible the development of measurement systems that allow electric networks to be fully observable only with the use of phasor measurements. The literature presents some techniques that are able to consider simultaneously phasor measurements and conventional measurements in the state estimation process. In this work, the main state estimators of one and two phases available in the literature are presented and evaluated. Moreover, the traditional state estimator was used to compare results. In the first part of evaluation process, the quality of state estimators in the presence of perfect measurements and measurements containing random gaussian errors is analyzed. Gaussian random errors are typical of measurement processes and therefore, they are considered acceptable. In the second part, the state estimators are evaluated in the presence of measurements containing random gaussian errors and gross errors. Gross errors, unlike the gaussian errors, are not acceptable and must be removed during the state estimation process. In order to make the analysis more comprehensive and realistic, the three main mechanisms of standard deviation calculation that are proposed in the literature are considered in state estimators evaluation. So, scenarios with fixed standard deviations, standard deviations calculated according to measurement values and standard deviations calculated according to measurements values and the full scale of meters are considered. For the detection and identification of gross errors, the large normalized residual test (LNR) is adopted, since this test is more reliable, robust and acceptable for this purpose. Tests in IEEE14 network are carried out. This network is widely adopted in theses and articles specialized in the field which facilitates the comparison of results / Mestrado / Energia Eletrica / Mestre em Engenharia Elétrica

Dans le contexte des systèmes dynamiques, cette thèse développe des techniques d'estimation d'état ensemblistes pour différentes classes de systèmes. On considère pour cela le cas d'un système standard linéaire invariant dans le temps soumis à des perturbations, des bruits de mesure et des incertitudes inconnus, mais bornés. Dans une première étape, une technique d'estimation d'état ellipsoïdale est étendue, puis appliquée sur un modèle d'octorotor utilisé dans un contexte radar. Une extension de cette approche ellipsoïdale d'estimation d'état est proposée pour des systèmes descripteurs. Dans la deuxième partie, nous proposons une méthode fondée sur la minimisation du P-rayon d'un zonotope, appliquée à un modèle d'octorotor. Cette méthode est ensuite étendue pour traiter les systèmes affines par morceaux. Dans la continuité des approches précédentes, un nouveau filtre de Kalman sous contraintes zonotopiques est proposé dans la dernière partie de cette thèse. En utilisant la forme duale d'un problème d'optimisation, l'algorithme projette l'état sur un zonotope qui forme l'enveloppe de l'ensemble des contraintes auxquelles l'état est soumis. La complexité de l'algorithme est ensuite améliorée en remplaçant le zonotope initial par une forme réduite en limitant son nombre de générateurs. / In the context of dynamical systems, this thesis focuses on the development of robust set-membership state estimation procedures for different classes of systems. We consider the case of standard linear time-invariant systems, subject to unknown but bounded perturbations and measurement noises. The first part of this thesis builds upon previous results on ellipsoidal set-membership approaches. An extended ellipsoidal set-membership state estimation technique is applied to a model of an octorotor used for radar applications. Then, an extension of this ellipsoidal state estimation approach is proposed for descriptor systems. In the second part, we propose a state estimation technique based on the minimization of the P-radius of a zonotope, applied to the same model of the octorotor. This approach is further extended to deal with piecewise affine systems. In the continuity of the previous approaches, a new zonotopic constrained Kalman filter is proposed in the last part of this thesis. By solving a dual form of an optimization problem, the algorithm projects the state on a zonotope forming the envelope of the set of constraints that the state is subject to. Then, the computational complexity of the algorithm is improved by replacing the original possibly large-scale zonotope with a reduced form, by limiting its number of generators.

Dans cette thèse, nous aborderons deux aspects fondamentaux qui se posent dans les systèmes de commande modernes du fait de l'interaction entre des processus en temps continu et des dispositifs numériques: la synthèse de lois de commande en présence de quantificateurs et l'estimation d'état en présence de mesures sporadiques. Une des caractéristiques principales de cette thèse consiste également à proposer des méthodes constructives pour résoudre les problèmes envisagés. Plus précisément, pour répondre à cette exigence, nous allons nous tourner vers une approche basée sur les inégalités matricielles linéaires (LMI). Dans la première partie de la thèse, nous proposons un ensemble d'outils constructifs basés sur une approche LMI, pour l'analyse et la conception de systèmes de commande quantifiés impliquant des modèles et des correcteurs linéaires. L'approche est basée sur l'utilisation des inclusions différentielles qui permet de modéliser finement le comportement de la boucle fermée et ainsi d'obtenir des résultats intéressants. Dans la seconde partie de la thèse, inspirés par certains schémas d'observation classiques présentés dans la littérature, nous proposons deux observateurs pour l'estimation de l'état d'un système linéaire en présence de mesures sporadiques, c'est-à-dire prenant en compte la nature discrète des mesures disponibles. De plus, en se basant sur une des deux solutions présentées, une architecture de commande basée observateur est proposée afin de stabiliser asymptotiquement un système linéaire en présence à la fois de mesures sporadiques et d'un accès intermittent à l'entrée de commande du système. / In this dissertation, two fundamental aspects arising in modern engineered control systems will be addressed:On the one hand, the presence of quantization in standard control loops. On the other hand, the state estimation in the presence of sporadic available measurements. These two aspects are addressed in two different parts. One of the main feature of this thesis consists of striving to derive computer-aided tools for the solution to the considered problems. Specifically, to meet this requirement, we revolve on a linear matrix inequalities (LMIs) approach. In the first part, we propose a set of LMI-based constructive Lyapunov-based tools for the analysis and the design of quantized control systems involving linear plants and linear controllers. The entire treatment revolves on the use of differential inclusions as modeling tools, and on stabilization of compact sets as a stability notion. In the second part of the thesis, inspired by some of the classical observation schemes presented in the literature of sampled-data observers, we propose two observers to exponentially estimate the state of a linear system in the presence of sporadic measurements. In addition, building upon one of the two observers, an observer-based controller architecture is proposed to asymptotically stabilize a linear plant in the presence of sporadic measurements and intermittent input access.