Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Monitoring of process control systems is extremely important for industries to ensure the quality of the product and the safety of the process. Predictive controllers, also known by MPC (Model Predictive Control), usually has a well performance initially. However, after a period, many factors contribute to the deterioration of its performance. This highlights the importance of monitoring the MPC control systems. In this work, tools based on multivariate statistical methods are discussed and applied to the problem of monitoring and Performance Assessment of predictive controllers. The methods presented here are: PCA (Principal Component Analysis) and ICA (Independent Component Analysis). Both are techniques that use data collected directly from the process. The first is widely used in Performance Assessment of predictive controllers. The second is a more recent technique that has arisen, mainly in order to be used in fault detection systems. The analyzes are made when applied in simulated processes characteristic of the petrochemical industry operating under MPC control. / O monitoramento de sistemas de controle de processos é extremamente importante no que diz respeito às indústrias, para garantir a qualidade do que é produzido e a segurança do processo. Os controladores preditivos, também conhecidos pela sigla em inglês MPC (Model Predictive Control), costumam ter um bom desempenho inicialmente. Entretanto, após um certo período, muitos fatores contribuem para a deterioração de seu desempenho. Isto evidencia a importância do monitoramento dos sistemas de controle MPC. Neste trabalho aborda-se ferramentas, baseada em métodos estatísticos multivariados, aplicados ao problema de monitoramento e avaliação de desempenho de controladores preditivos. Os métodos aqui apresentados são: o PCA (Análise por componentes principais) e o ICA (Análise por componentes independentes). Ambas são técnicas que utilizam dados coletados diretamente do processo. O primeiro é largamente utilizado na avaliação de desempenho de controladores preditivos. Já o segundo, é uma técnica mais recente que surgiu, principalmente, com o intuito de ser utilizado em sistemas de detecção de falhas. As análises são feitas quando aplicadas em processos simulados característicos da indústria petroquímica operando sob controle MPC.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / The data reconciliation in process control is extremely important regarding the industries because from this it is possible to obtain a greater efficiency in the performance in industrial process control meshes aiming at a lower cost and a higher quality of the product. In this work we approach data estimation techniques for the implementation of an online dynamic data reconciliation system in order to reduce the noise and the measurement uncertainties that are submitted in the process variables. The techniques used here are: the Kalman Filter, the Preditor-Corrector DDR Algorithm, the Moving Horizon Estimator (MHE) and the Constrained Extended Kalman Filter (CEKF). The analysis is performed by applying the dynamic data reconciliation system in a simulated process, characteristic of the chemical industry, operating under MPC (Model Predictive Control). The performance of the MPC controller is also enhanced by the use of the reconciled data in the feedback control loop. / A reconciliação de dados em controle de processos é extremamente importante no que diz respeito às indústrias, pois a partir dessa é possível obter uma maior eficiência no desempenho em malhas de controle de processos industriais visando à minimização dos custos e maximizando a qualidade do produto. Neste trabalho abordam-se técnicas de estimação de dados para a implementação de um sistema de reconciliação dinâmica de dados on-line a fim de reduzir os ruídos e as incertezas de medições a que estão submetidas às variáveis do processo. As técnicas aqui empregadas são: o Filtro de Kalman, o Algoritmo DDR Preditor-Corretor, o Estimador de Horizonte Móvel (MHE) e o Filtro de Kalman Estendido com Restrições (CEKF). As análises são efetuadas aplicando o sistema de reconciliação dinâmica de dados em um processo simulado, característico da indústria química, operando sob controle preditivo (MPC). Também é efetuado o aprimoramento no desempenho do controlador MPC utilizando os dados reconciliados na malha de realimentação do controlador.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Model predictive control (MPC) strategies have become the standard for advanced control applications in the process industry. Significant benefits are generated from the MPC's capacity to ensure that the plant operates within its constraints more profitably. However, like any controller, after some time under operation, MPCs rarely function as when they were initially designed. A large percentage of performance degradation of MPC is associated with the deterioration of model that controller uses to predict process outputs and calculate inputs. The objective of the present work is implementation of mathematical methods that can be used to detect model-plant mismatch in linear and nonlinear MPC systems. In this work, techniques based on cross correlation, partial correlation and mutual information are implemented and tested by numerical simulation in case studies characteristic of the petrochemical industry, represented by linear and nonlinear models, operating under MPC control. The results obtained through the applying the techniques are analyzed and compared as to their efficiency is not intended to offer their potential for real industrial applications. / Estratégias de controle preditivo (MPC) têm-se tornado o padrão para aplicações de controle avançado na indústria de processos. Os benefícios significativos são gerados a partir da habilidade do controlador MPC de assegurar que a planta opere dentro das restrições de forma mais lucrativa. Porém, como todo controlador, depois de algum tempo em operação, os MPCs raramente funcionam como quando foram inicialmente projetados. Uma grande porcentagem da degradação do desempenho dos controladores MPC está associada à deterioração do modelo que o controlador usa para fazer a predição das saídas do processo e calcular as entradas. O objetivo do presente trabalho é a implementação de métodos matemáticos que possam ser utilizados para a detecção de erros planta-modelo em sistemas de controle MPC lineares e não lineares. Neste trabalho, técnicas baseadas em correlação cruzada, correlação parcial e informação mútua são implementadas e testadas por simulação numérica em estudos de caso característicos da indústria petroquímica, representados por modelos lineares e não lineares, operando sob controle MPC. Os resultados obtidos através da aplicação das técnicas são analisados e comparados quanto à sua eficiência no objetivo proposto avaliando seu potencial para aplicações industriais reais.

Submitted by Luciana Ferreira (lucgeral@gmail.com) on 2017-10-06T12:04:38Z No. of bitstreams: 2 Dissertação - Lázaro Rubens Araújo Pinto - 2017.pdf: 16272237 bytes, checksum: 9f3d1693d0282cbf28a35702c1a9cf17 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2017-10-06T12:04:51Z (GMT) No. of bitstreams: 2 Dissertação - Lázaro Rubens Araújo Pinto - 2017.pdf: 16272237 bytes, checksum: 9f3d1693d0282cbf28a35702c1a9cf17 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Made available in DSpace on 2017-10-06T12:04:52Z (GMT). No. of bitstreams: 2 Dissertação - Lázaro Rubens Araújo Pinto - 2017.pdf: 16272237 bytes, checksum: 9f3d1693d0282cbf28a35702c1a9cf17 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2017-08-07 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / This work proposes an analysis of the closed-loop operation of a distributed generation system connected to a three-phase four-wire AC power system by a three-phase four-leg inverter. The modulation strategy adopted for this inverter is determined by the Model Predictive Control (MPC) method. Reasons for the use of this topology and the control method are discussed. The chosen structure for the MPC predictive control aims to supply of active and reactive power by the inverter in order to maximize the power factor measured by the four-wire AC source (Y-n). Four possible cases of operation are reported for the analysis of the proposal in the three-phase four-leg inverter and other two cases for a three-phase full-bridge inverter (3F) for comparison purposes. Based on the monitoring of the active powers involved and the other variables of interest, it was possible to exemplify the achievement of the proposed objectives and validate the adopted method, at least under modeling and software simulation level. / Este trabalho propõe uma análise da operação em malha fechada de um sistema de geração distribuída para conexão em um sistema trifásico a quatro fios, a partir do controle de um inversor trifásico a quatro braços. A estratégia de modulação adotada para este inversor é determinada pelo método de controle preditivo baseado em modelo (MPC - Model Predictive Control). Justificativas para a utilização dessa topologia e do método de controle são discutidas. A estrutura escolhida para o controle preditivo MPC objetiva o fornecimento de potência ativa e reativa pelo inversor tendo em vista a maximização do fator de potência medido a partir da fonte pela fonte CA a quatro fios (Y-n). Quatro casos são abordados para análise da proposta no inversor trifásico a quatro braços e dois casos para um inversor trifásico em ponte completa (3F) para efeito de comparação. Baseando-se no monitoramento das potências elétricas envolvidas e nas demais grandezas de interesse, foi possível concluir que os resultados de simulações exemplificam o alcance dos objetivos propostos e validam o método adotado, pelo menos em relação à modelagem matemática empregada e à simulação do sistema de geração proposto em ambiente computacional.

Le réglage de la tension et la gestion de la puissance réactive est d’une importance capitale pour le bon fonctionnement du système électrique. Les réseaux de distribution connaissent des modifications profondes qui sont tant techniques avec l’insertion de la production décentralisée ou l’enfouissement des lignes aériennes, que réglementaires avec l’entrée en vigueur des codes de réseaux Européens. Aussi, des alternatives aux réglages traditionnels de tension et la mise en place de contrôle des échanges de puissance réactive à l’interface réseau de transport/distribution peuvent être développées. Dans le cadre de ces travaux de doctorat, une solution basée sur l’amélioration du réglage temps réel des réseaux de distribution ainsi que la mise en place d’une gestion conjointe de la puissance réactive entre les gestionnaires du réseau de transport et distribution a été proposée. Ce réglage temps réel est basé sur une méthode de commande prédictive, et s’appuie notamment sur le régleur en charge ou les productions décentralisées pour réaliser un réglage de la tension au sein d’un réseau de distribution et contrôler les échanges de puissance réactive. Les références de puissance réactive à atteindre à l’interface entre réseau de distribution et de transport sont déterminées par le gestionnaire du réseau de transport pour ses propres besoins et en connaissance des réserves de puissance réactive disponible depuis les réseaux de distribution. Par rapport à la littérature, notre démarche a pour originalité de prendre en compte les problèmes à l’interface des réseaux de distribution et de transport et démontre l’intérêt de mener des études conjointes entre gestionnaires / Voltage and reactive power control are of paramount importance to ensure safe and reliable operation of the power system. Distribution grids are undergoing major changes, namely the insertion of distributed generation and the replacement of overhead lines. Along with these physical evolutions, new distribution networks should comply with the requirements of the European Grid Codes on the reactive power exchange at the HV/MV interface. To handle these new operational concerns, alternative solutions to the traditional voltage and reactive power control can be found. In our work, a scheme based on the evolution of the real-time Volt Var Control (VVC) of distribution networks and a joint coordination for the reactive power management of a HV system has been proposed. The real-time VVC of MV grids is based on a predictive control method. This control uses in a coordinated manner the on load tap changer, the distributed generation and the capacitor banks to enforce a suitable MV voltage profile and an appropriate HV/MV reactive power exchange. The reactive power targets at the HV/MV interface are determined by the Transmission System Operator for its own requirements but considering the true MV reactive power reserve. Compared to the literature, the novelty of our approach consists in considering the concerns at the HV/MV system interface. Next our works have shown the relevance of performing joint transmission and distribution system operators analysis.

Un bras de chargement de gaz est une structure articulée dans laquelle du méthane peut s’écouler à température cryogénique. En haute mer, ces bras sont installés sur le pont de navires-usines et se connectent à des méthaniers pour leur transférer du gaz. En raison de problèmes de sécurité et de performances, il est souhaité que le bras de chargement soit robotisé pour qu’il se connecte automatiquement. Cette thèse a pour objectif l‘automatisation de la connexion. Cette opération nécessite un pilotage de grande précision vis à vie de la taille du bras. Pour cette raison le bras est d’abord étalonné pour augmenter sa précision statique. Ensuite, des analyses modales expérimentales mettent en évidence l’importante souplesse de la structure des bras de chargement. Pour cette raison un générateur de trajectoires « douces », à jerk limité, est développé afin de piloter le bras sans le faire vibrer. Enfin, un système de compensation actif visant à compenser les mouvements relatifs des deux navires est mis en place. Cette compensation combine la génération de trajectoires douces avec une composante prédictive basée sur des réseaux de neurones. Cette dernière permet de prédire et d’anticiper les mouvements des navires sur l’océan, afin d’annuler tout retard dans la compensation. Finalement, cette thèse présente la première connexion automatique d’un bras de chargement, et démontre la validité de cette approche. / Marine loading arms are articulated structures that transfer liquefied gas between two vessels. The flanging operation of the loading arm to the receiving tanker is very sensitive. This thesis aims to robotize a loading arm so it can flange automatically. The required accuracy for the connection is very high. A calibration procedure is thus proposed to increase the accuracy of loading arms. Moreover a jerk-limited trajectory generator is developed to smoothly drive the arm without inducing oscillation. This element is important because the structures of loading arms have a very low stiffness and easily oscillate, as highlighted by modal analyses.A predictive active compensation algorithm is developed to track without delay the relative motion between the two vessels. This algorithm relies on an artificial neural network able to predict the evolution of this relative motion. Finally this thesis presents the first automatic connection of an offshore loading arm. The success of the final tests validate the feasibility the automatic connection and the validity of this approach.

The interest in autonomous vehicles has never been higher and there are several components that need to function for a vehicle to be fully autonomous; one of which is the ability to perform a parking at the end of a mission. The objective of this thesis work is to develop and implement an automatic parking system (APS) for a heavy-duty vehicle (HDV). A delimitation in this thesis work is that the parking lot has a known structure and the HDV is a truck without any trailer and access to more computational power and sensors than today's commercial trucks. An automatic system for searching the parking lot has been developed which updates an occupancy grid map (OGM) based on measurements from GPS and LIDAR sensors mounted on the truck. Based on the OGM and the known structure of the parking lot, the state of the parking spots is determined and a path can be computed between the current and desired position. Based on a kinematic model of the HDV, a gain-scheduled linear quadratic (LQ) controller with feedforward action is developed. The controller's objective is to stabilize the lateral error dynamics of the system around a precomputed path. The LQ controller explicitly takes into account that there exist an input delay in the system. Due to minor complications with the precomputed path the LQ controller causes the steering wheel turn too rapidly which makes the backup driver nervous. To limit these rapid changes of the steering wheel a controller based on model predictive control (MPC) is developed with the goal of making the steering wheel behave more human-like. A constraint for maximum allowed changes of the controller output is added to the MPC formulation as well as physical restrictions and the resulting MPC controller is smoother and more human-like, but due to computational limitations the controller turns out less effective than desired. Development and testing of the two controllers are evaluated in three different environments of varying complexity; the simplest simulation environment contains a basic vehicle model and serves as a proof of concept environment, the second simulation environment uses a more realistic vehicle model and finally the controllers are evaluated on a full-scale HDV. Finally, system tests of the APS are performed and the HDV successfully parks with the LQ controller as well as the MPC controller. The concept of a self-parking HDV has been demonstrated even though more tuning and development needs to be done before the proposed APS can be used in a commercial HDV.

Dans un objectif conjoint d'identification paramétrique en ligne, les méthodes développées dans cette thèse permettent de concevoir en ligne et en boucle fermée les entrées optimales qui enrichissent les informations contenues dans l'expérience en cours. Ces méthodes reposent sur des mesures en temps réel du procédé, sur un modèle dynamique non linéaire (ou linéaire) multi-variable choisi du procédé, sur un modèle de sensibilité des mesures par rapport aux paramètres à estimer et sur un observateur non linéaire. L'analyse de l'observabilité et des techniques de commande prédictive permettent de définir la commande optimale qui est déterminée en ligne par optimisation sous contraintes. Des aspects de stabilisation sont également étudiés (via un apport de contraintes fictives ou via une technique de Lyapunov). Enfin, une loi de commande explicite pour le cas particulier du système d'ordre un est développée. Des exemples illustratifs sont traités via le logiciel ODOE4OPE : un bioréacteur, un réacteur continu parfaitement agité et une aile delta. Ces exemples permettent de voir que l'estimation des paramètres peut être réalisée avec une bonne précision, et à moindre coût expérimental en une expérience / For online parameter identification, the developed methods here allow to design online and in closed loop optimal inputs that enrich the information in the current experience. These methods are based on real-time measurements of the process, on a dynamic nonlinear (or linear) multi-variable model, on a sensitivity model of measurements with respect to the parameters to be estimated and a nonlinear observer. Analysis of observability and predictive control techniques are used to define the optimal control which is determined online by constrained optimization. Stabilization aspects are also studied (by adding fictitious constraints or by a Lyapunov technique). Finally, for the particular case of a first order linear system, the explicit control law is developed. Illustrative examples are processed via the ODOE4OPE software : a bio-reactor, a continuous stirred tank reactor and a delta wing. These examples help to see that the parameter estimation can be performed with good accuracy in a single and less costly experiment

Lors de l’installation d’une pompe à chaleur (PAC) double service (chauffage et eau chaude sanitaire (ECS)) dans le secteur résidentiel, la phase de mise en service est délicate : les paramètres à renseigner sont nombreux et non triviaux. Or, le bon fonctionnement de la PAC est très sensible à la qualité de cette étape. Quelques mauvais réglages peuvent entraîner un fonctionnement non optimal, voire un dysfonctionnement important (confort mal assuré). L’objectif de la thèse est de mettre au point une méthode de paramétrage auto-adaptatif modifiant les valeurs de « sortie d’usine » des paramètres pour les adapter au réseau d'émetteurs, à la réponse thermique du bâtiment et aux habitudes chauffage et de consommation d'ECS des occupants. Les paramètres doivent être adaptés à partir des capteurs déjà en place sur la PAC.Le travail de thèse porte d'abord sur l’identification des paramètres clés de la PAC, ceux qui influencent le plus la consommation. On en déduit quelques fonctions à optimiser : ajustement de la loi d'eau, méthode de choix des meilleures séquences de production d'ECS, anticipation de la relance du chauffage. L'objectif est de maximiser le coefficient de performance et de minimiser le recours aux appoints électriques tout en garantissant le confort. Ces objectifs sont atteints en développant des algorithmes de contrôle optimisé. Des modèles neuronaux de prévision de la réponse thermique du bâtiment, du stock d’ECS et des performances de la PAC ont été développés pour ce contrôle optimisé. Les modèles et algorithmes développés ont été validés numériquement et les performances de la PAC comparées à celles avec contrôle classique sans auto-paramétrage. Les solutions proposées ont été appliquées et testées durant une saison sur une PAC réelle sur un banc d'essai semi-virtuel (climat réel et bâtiment virtuel). / Setting control parameters of residential double service heat pumps at the time of installation and commissioning is a delicate matter. Indeed, some parameters are not trivial, there are many to be adjusted and the heat pump operations are quite sensitive to the parameters' values. Poor parameterization can lead to suboptimal heat pump operation or even to important dysfunction (harming thermal comfort).Hence, this thesis aims to develop a method for the heat pump to self-adapt the value of its control parameters. The heat pump should modify if required the "default" settings in order to adapt them to the heat emitters, to the building thermal response and to the occupancy (in terms of thermal comfort and DHW needs). For industrial reasons, this method should use on-board sensors.First, the thesis focuses on identifying the key parameters of the heat pump control, i.e. those with greatest influence on the consumption. This leads to the functions which have to be optimized: heating curve adjustment, time of DHW generation, heating setback anticipation. The objective is to maximize the coefficient of performance and minimize the use of electrical back-ups while ensuring comfort. This is achieved by developing optimized control algorithms. Thanks to forecasts models, based on neural networks, we are able to predict on a short term horizon the building thermal response, the DHW availability and the heat pump performances. The developed models and algorithms have been validated through numerical simulations, and we have evaluated the heat pump performances in comparison to a classic control. The proposed solutions were applied and tested during a heating season on a real heat pump installed in a semi-virtual test bench (real weather and virtual building).

La problématique étudiée dans cette thèse concerne le guidage en formation d’une flotte de robots mobiles en environnement naturel. L’objectif poursuivi par les robots est de suivre une trajectoire connue (totalement ou partiellement) en se coordonnant avec les autres robots pour maintenir une formation décrite comme un ensemble de distances désirées entre les véhicules. Le contexte d’évolution en environnement naturel doit être pris en compte par les effets qu’il induit sur le déplacement des robots. En effet, les conditions d’adhérence sont variables et créent des glissements significatifs des roues sur le sol. Ces glissements n’étant pas directement mesurables, un observateur est mis en place, permettant d’obtenir une estimation de leur valeur. Les glissements sont alors intégrés au modèle d’évolution, décrivant ainsi un modèle cinématique étendu. En s’appuyant sur ce modèle, des lois de commande adaptatives sur l’angle de braquage et la vitesse d’avance d’un robot sont alors conçues indépendamment, asservissant respectivement son écart latéral à la trajectoire et l’interdistance curviligne de ce robot à une cible. Dans un second temps, ces lois de commande sont enrichies par un algorithme prédictif, permettant de prendre en compte le comportement de réponse des actionneurs et ainsi d’éviter les erreurs conséquentes aux retards de la réponse du système aux commandes. À partir de la loi de commande élémentaire en vitesse permettant d’assurer un asservissement précis d’un robot par rapport à une cible, une stratégie de commande globale au niveau de la flotte est établie. Celle-ci décline l’objectif de maintien de la formation en consigne d’asservissement désiré pour chaque robot. La stratégie de commande bidirectionnelle conçue stipule que chaque robot définit deux cibles que sont le robot immédiatement précédent et le robot immédiatement suivant dans la formation. La commande de vitesse de chaque robot de la formation est obtenue par une combinaison linéaire des vitesses calculées par la commande élémentaire par rapport à chacune des cibles. L’utilisation de coefficients de combinaison constants au sein de la flotte permet de prouver la stabilité de la commande en formation, puis la définition de coefficients variables est envisagée pour adapter en temps réel le comportement de la flotte. La formation peut en effet être amenée à évoluer, notamment en fonction des impératifs de sécurisation des véhicules. Pour répondre à ce besoin, chaque robot estime en temps réel une distance d’arrêt minimale en cas d’urgence et des trajectoires d’urgence pour l’évitement du robot précédent. D’après la configuration de la formation et les comportements d’urgence calculés, les distances désirées au sein de la flotte peuvent alors être modifiées en ligne afin de décrire une configuration sûre de la formation. / This thesis focuses on the issue of the control of a formation of wheeled mobile robots travelling in off-road conditions. The goal of the application is to follow a reference trajectory (entirely or partially) known beforehand. Each robot of the fleet has to track this trajectory while coordinating its motion with the other robots in order to maintain a formation described as a set of desired distances between vehicles. The off-road context has to be considered thoroughly as it creates perturbations in the motion of the robots. The contact of the tire on an irregular and slippery ground induces significant slipping and skidding. These phenomena are hardly measurable with direct sensors, therefore an observer is set up in order to get an estimation of their value. The skidding effect is included in the evolution of each robot as a side-slip angle, thus creating an extended kinematic model of evolution. From this model, adaptive control laws on steering angle and velocity for each robot are designed independently. These permit to control respectively the lateral distance to the trajectory and the curvilinear interdistance of the robot to a target. Predictive control techniques lead then to extend these control laws in order to account for the actuators behavior so that positioning errors due to the delay of the robot response to the commands are cancelled. The elementary control law on the velocity control ensures an accurate longitudinal positioning of a robot with respect to a target. It serves as a base for a global fleet control strategy which declines the overall formation maintaining goal in local positioning objective for each robot. A bidirectionnal control strategy is designed, in which each robot defines 2 targets, the immediate preceding and following robot in the fleet. The velocity control of a robot is finally defined as a linear combination of the two velocity commands obtained by the elementary control law for each target. The linear combination parameters are investigated, first defining constant parameters for which the stability of the formation is proved through Lyapunov techniques, then considering the effect of variable coefficients in order to adapt in real time the overall behavior of the formation. The formation configuration can indeed be prone to evolve, for application purposes and to guarantee the security of the robots. To fulfill this latter requirement, each robot of the fleet estimates in real time a minimal stopping distance in case of emergency and two avoidance trajectories to get around the preceding vehicle if this one suddenly stops. Given the initial configuration of the formation and the emergency behaviors calculated, the desired distances between the robots can be adapted so that the new configuration thus described ensures the security of each and every robot of the formation against potential collisions.