The Linear Quadratic Regulator (LQR) is an optimal control approach which aims to drive states of a linear system to its origin through the minimization of a quadratic cost functional. Such an approach has been widely successful for both theoretical and practical applications. However, when such controllers are subject to uncertainties, optimal closed-loop performance cannot be obtained since robustness properties are no longer guaranteed. Guaranteed Cost Controllers (GCC) presents robust asymptotic stability and provides a guaranteed upper bound to a quadratic cost function. Such method addresses the lack of performance guarantees of the LQR. Meanwhile, Model Predictive Control (MPC) is a class of optimization-based control algorithms that use an explicit model of the controlled system to predict its future states. The MPC can be as a generalization of the LQR for constrained linear systems. Therefore, it equally suffers from a lack of robustness guarantees when the system is subject to uncertainties. Robust MPC (RMPC) approaches were proposed to address MPCs poor closed-loop performance subject to uncertainties. Its objective is to obtain a control input sequence that simultaneously minimizes a cost function and guarantees the feasibility of system states and control inputs, for a system subject to the worst-case disturbance within an uncertainty set. Autonomous vehicles have gained increasing interest from both the industry and research communities in recent years. An essential aspect in the design of automotive control systems is to ensure the controller is stable and has acceptable performance within the entire operational envelope which it is designed to operate. In the case of autonomous vehicles, where there is no human driver as a fallback, it is of utmost importance to ensure the safe operations of the control system and its capability to avoid saturating the handling limits of the vehicle. In this thesis, we propose Guaranteed Cost Controller approaches for both unconstrained and constrained linear systems subject to multiplicative structured norm-bounded uncertainties and present the application of such a controller to the lateral control problem of autonomous vehicles up to the tire saturation limits. / O Regulador Quadrático Linear (Linear Quadratic Regulator, LQR) é uma abordagem de controle ótimo que visa conduzir estados de um sistema linear à sua origem através da minimização de um custo funcional quadrático. Tal abordagem tem sido amplamente bem sucedida para aplicações teóricas e práticas. No entanto, não é possível obter o desempenho ótimo de malha fechada quando esses controladores são sujeitos a incertezas no sistema em decorrência de suas propriedades de robustez não serem garantidas. Controladores de Custo Garantido (Guaranteed Cost Control, GCC) visam abordar a falta de garantia de desempenho do LQR, neste caso. Esses controladores apresentam estabilidade assintótica robusta e fornecem um custo garantido de pior caso para uma função de custo quadrático. O Controle Preditivo de Modelo (Model Predictive Control, MPC) é uma classe de algoritmos de controle baseados em otimização que usa um modelo explícito do sistema controlado para prever seus estados futuros. Uma possível interpretação do MPC é uma generalização do LQR para sistemas lineares com restrições de estado e entrada de controle. Portanto, essa abordagem sofre igualmente da falta de garantias de robustez quando o sistema é sujeito a incertezas. As abordagens de MPC Robustas (Robust MPC, RMPC) foram propostas para abordar o desempenho de malha fechada do MPC sujeito a incertezas no sistema. Seu objetivo é obter uma sequência de entrada de controle que minimize simultaneamente uma função de custo e garanta que os estados do sistema e as entradas de controle estão contidos dentro das restrições para um sistema sujeito à pior das perturbações dentro de um conjunto admissível de incertezas. Pesquisas voltadas para veículos autônomos ganharam crescente interesse nos últimos anos, tanto da indústria automobilística quanto da comunidade acadêmica. Um aspecto essencial no projeto de sistemas de controle automotivo é a garantia de estabilidade e desempenho do controlador dentro de todo o envelope operacional ao qual ele foi projetado para operar. No caso de veículos autônomos, onde não há motoristas humanos para lidar com casos de falha do sistema, é de suma importância assegurar as operações seguras do sistema de controle e sua capacidade de evitar a saturação dos limites de manuseio do veículo. Nesta tese, propomos abordagens GCC para sistemas lineares restritos e irrestritos, sujeitos a incertezas estruturadas contidas por norma e apresentamos a aplicação de tais controladores ao problema de controle lateral de veículos autônomos até os limites de saturação dos pneus.

Garantir le confort thermique des bâtiments est directement lié à la consommation d'énergie. Dans les zones tropicales, les systèmes de refroidissement représentent l'un des postes les plus gourmands en énergie. Afin de réduire la consommation d'énergie mondiale, il est primordial d'améliorer l'efficacité de ces systèmes ou bien de développer de nouvelles méthodes de production de froid. Une installation de refroidissement solaire basé sur le cycle à absorption est une alternative pour réduire les émissions de gaz à effet de serre et la consommation d'électricité. Contrairement aux systèmes classiques de refroidissement à compression mécanique, la production de froid par absorption est un système complexe composé de plusieurs composants comme des panneaux solaires, un ballon de stockage, une tour de refroidissement et une machine à absorption. Outre le dimensionnement des composants, ce système complexe nécessite des actions de contrôle pour être efficace parce que la coordination entre le stockage d'eau chaude, la production et la consommation du froid est nécessaire. Le but de cette thèse est de proposer une structure producteur-consommateur d'énergie basée sur la commande prédictive (MPC). Le système de refroidissement par absorption solaire est considéré comme faisant partie de ce système de production-consommation d'énergie, le système de stockage d'eau chaude est le producteur et la machine à absorption qui distribue de l'eau froide au bâtiment est l'un des consommateurs. Pour que la structure de commande soit modulaire, la coordination entre les sous-systèmes est réalisée en utilisant une approche de partitionnement où des contrôleurs prédictifs locaux sont conçus pour chacun des sous-systèmes. Les contrôleurs des consommateurs calculent un ensemble de profils de demande d'énergie. Ces profils sont ensuite envoyés au contrôleur du producteur qui sélectionne le profil qui minimise le coût global. Dans une première partie, l'approche proposée est testée sur un modèle linéaire simplifié composé d'un producteur et plusieurs consommateurs. Dans une deuxième partie, un cas plus complexe est étudié. Un modèle simplifié d'un système de refroidissement à absorption est évaluée en utilisant l'outil de simulation TRNSYS. Le modèle de production n'est plus linéaire, il est décrit par un modèle non linéaire hybride qui augmente la complexité du problème d'optimisation. Les résultats des simulations montrent que la sous-optimalité induite par la méthode est faible. De plus, la performance de l'approche atteint les objectifs de commande tout en respectant les contraintes. / To guarantee thermal comfort in buildings is directly related to energy consumption. In tropical climates, cooling systems for buildings represent one of the largest energy consumers. Therefore, as energy consumption is a major concern around the world, it is important to improve the systems efficiency or seeking new methods of cooling production. A solar cooling installation based on the absorption cycle is an alternative to mitigate greenhouse gas emissions and electricity consumption. In contrast to conventional vapor-compression based cooling systems, the absorption cooling production involves a complex system composed of several components as collector panel, storage tank, cooling tower and absorption chiller. Besides the sizing of the components, this complex system requires control actions to be efficient as a coordination between hot water storage, cooling water production and consumption is necessary. The aim of this research is to propose a management approach for a production-consumption energy system based on Model Predictive Control (MPC). The solar absorption cooling system is seen as part of this production-consumption energy system where the hot water storage system is the producer and the chiller-building system is one of the consumers. In order to provide modularity to the control structure, the coordination between the subsystems is achieved by using a partitioning approach where local predictive controllers are developed for each of the subsystems. The consumer controllers compute a set of energy demand profiles sent to the producer controller which selects the profile that better minimize the global optimization cost. In a first part, the proposed approach is tested on a simplified linear model composed of one producer and several consumers. In a second part, a more complex case is studied. A simplified model of an absorption cooling system is evaluated using the simulation tool TRNSYS. The producer model is no longer linear, instead it is described by a nonlinear hybrid model which increases the complexity of the optimization problem. The simulations results show that the suboptimality induced by the method is low and the control strategy fulfills the objectives and constraints while giving good performances.

En comparaison avec leurs homologues sériels, les robots parallèles possèdent de nombreux avantages : rigidité, temps de cycle et précision de positionnement. La faible taille de leur espace de travail opérationnel est cependant un inconvénient qui empêche leur développement. L’analyse cinématique décrit la division de l’espace de travail total en aspects, séparés entre eux par des singularités de Type 2. Parmi les solutions d’élargissement de l’espace de travail opérationnel issues de la littérature, que sont comparées entre elles, nous retiendrons la traversée des singularités grâce à une génération de trajectoire et une loi de commande dédiées. Cette solution est cependant sujette aux échecs de traversée et ne permet pas de certifier la réussite de l’opération. La première partie du travail consiste donc à développer un outil permettant de détecter le résultat des traversées. De tels algorithmes n’existent pas dans la littérature ; en effet, les solveurs du problème géométrique direct ne peuvent pas donner de résultats appropriés à proximité des singularités. Cependant les méthodes ensemblistes proposent une manière intéressante de tenir compte de la cinématique du robot. Nous nous basons sur cette considération pour développer un algorithme réalisant le suivi de la pose ainsi que de la vitesse de l’effecteur du robot. En simulation, nous démontrons sa capacité à détecter les changements de mode d’assemblage et son utilité pour générer des trajectoires permettant la traversée. La deuxième problématique consiste à améliorer le suivi de trajectoire par l’utilisation de techniques de commande avancée. Une revue de littérature nous permet d’identifier la commande adaptative et la commande prédictive comme deux schémas intéressants pour notre application. La synthèse d’un loi de commande adaptative par des outils linéaires est proposée et complétée par la prédiction des paramètres dynamiques suivant la méthode Predcitive Functionnal Control. Les gains apportés par les lois de commande proposées sont évalués en simulation. Afin de valider ces contributions, elles sont implémentées sur un robot parallèle plan à deux degrés de liberté nommé DexTAR. Nécessaire à la mise en oeuvre de l’algorithme de détection du mode d’assemblage, un étalonnage géométrique du robot est réalisé et les paramètres estimés sont certifiés de manière ensembliste. La capacité de l’algorithme à statuer sur le mode d’assemblage final est ensuite évaluée sur des trajectoires réelles. Ces résultats sont comparés avec ceux obtenus en simulation. De plus, les lois de commande développées sont implémentées sur le robot DexTAR et testés dans des cas de figure réalistes comme les traversées multiples ou la saisie d’objets.Les propositions formulées dans ce manuscrit permettent de répondre à ces problématiques, afin de faciliter l’utilisation des méthodes d’agrandissement de l’espace de travail par traversée de singularité de Type 2. / Compared to their serial counterparts, parallel robots have the edge in terms of rigidity,cycle time and positioning precision. However, the reduced size of their operationalworkspace is a drawback that limits their use in the industry. Kinematic analysis explainshow the workspace is divided in aspects, separated from each others by so-called Type 2singularities. Among existing solutions for workspace enlargement, which are evaluatedin this thesis, we chose to work on a method based on singularity crossing. This can beachieved thanks to dedicated trajectory generators and control strategies. Yet, failuresin crossing can still happen and crossing success cannot be certified.In consequence, the first part of the thesis consists in the development of an algorithmable to state on the results of a crossing attempt. Such a tool does not exist inthe literature, since solvers for the forward kinematics of parallel robots diverge aroundsingularities. Nonetheless, interval methods allow to bypass this problem by trackingend-effector velocity alongside with its pose. The ability of the algorithm to detect assemblymode change is proven in simulation, and its usefulness for reliable trajectoryplanning is shown.In a second part, we seek to improve trajectory tracking through the use of advancedcontrol techniques. A review on those techniques showed adaptive control and predictivecontrol methods to be well-fitted to our application. Linear synthesis of articularadaptive control is proposed and then derived in order to predict dynamic parametersthanks to the Predictive Functional Control method. Efficiency of the proposed controllaws is evaluated in simulation.1In order to validate both contributions, algorithms and control laws are implementedon a 2-degree of freedom planar parallel robot named DexTAR. As it is mandatory forassembly mode detection, the kinematic calibration of the robot is completed from whichcertified geometric parameters are derived. Assembly mode detection is performed onreal trajectories and results are compared to those obtained in simulation. Moreover,adaptive and predictive control laws are tested in realistic cases of singularity crossingand object manipulation.Overall, proposed contributions answer the problems that were stated previously andform an improvement to the workspace enlargement method based on Type 2 singularitycrossing.

Dans l'industrie sidérurgique, les fours de réchauffage sont les plus grands consommateurs d'énergie après les hauts fourneaux. Réduire leur consommation énergétique est donc la préoccupation majeure de la commande des fours. Dans un four de réchauffage, des brames d'acier sont chauffées en traversant successivement plusieurs zones, de la température ambiante à un profil de température homogène de 1250 °C en sortie du four, avant d’être laminées dans les laminoirs à chaud. La température de brames est contrôlée par une structure de commande hiérarchisée à deux niveaux (niveau 1 et 2).L'objectif de ces travaux est d'améliorer la performance du chauffage et donc de réduire la consommation énergétique du four via une stratégie de commande prédictive distribuée hiérarchisée sur les deux niveaux de commande. Une approche de commande prédictive distribuée est tout d’abord développée pour le niveau 1 afin de suivre les consignes de température de zone, prenant en compte les couplages entre les zones et induisant une moindre complexité d’implantation par rapport à une approche centralisée. L’implantation industrielle a permis une amélioration significative de la précision du suivi de température et une réduction de la consommation d'énergie de 3%. Une deuxième étape propose l’élaboration de la commande prédictive hiérarchisée du niveau 2 afin, à partir de la consigne de température de brame, de déterminer les consignes de température optimales des zones en se fondant sur un modèle de transfert thermique du four. Les résultats de simulation, comparés aux données industrielles, montrent une réduction de la consommation énergétique de 5% et une meilleure qualité de chauffage des brames. L’approche précédente est enfin étendue pour prendre en compte et optimiser le cadencement des brames afin d’augmenter la productivité du four. La simulation montre une augmentation potentielle de productivité du four de 15 tonnes par heure tout en améliorant la qualité de chauffage des brames. / In steel industry, reheating furnaces are the biggest energy consumers after blast furnaces. As a result, reduction of energy consumption is the major concern of furnace control. In a walking-beam slab reheating furnace, steel slabs are heated by moving through successive zones from ambient temperature to a homogenous temperature profile of 1250°C at the furnace exit, to be rolled subsequently in the hot rolling mills. Temperature of slabs is controlled mainly by a two-level hierarchical structure, so called level 1 and level 2.The aim of this thesis is to improve the heating performance and consequently to reduce the energy consumption of the furnace by using hierarchical distributed model predictive control (MPC) strategy for both levels. In a first step, distributed model predictive controllers are developed for the level 1 in order to track zone temperature set-points. The distributed feature of the control law enables to consider coupling effects between zones while reducing the computation complexity compared to a complete centralized approach. The industrial results showed significant improvement on temperature tracking accuracy and an energy consumption reduction of 3%. In a second step, the hierarchical MPC is constructed for the level 2 in order to determine the optimal zones temperature setpoint from the slab temperature setpoint, based on a numerical heat transfer model of the furnace. The simulation results obtained with this strategy compared against industrial data show an energy consumption reduction of 5% and a better heating quality. The previous structure is finally extended to take into account and optimize the scheduling of the slabs within the MPC level 2 in order to increase productivity of the considered furnace. The simulation shows a potential increase of productivity of the furnace of 15 tons per hour while improving the slab heating quality.

Critical automated maneuvers for ships typically require a redundant set of thrusters. The motion control system hierarchy is commonly separated into several layers using a high-level motion controller and a thruster allocation (TA) algorithm. This allows for a modular design of the software where the high-level controller can be designed without comprehensive information on the thrusters, while detailed issues such as input saturation and rate limits are handled by the TA. However, for a certain set of thruster configurations this decoupling may result in poor control performance due to the limited knowledge in the high-level controller about the physical limitations of the ship and the behavior of the TA. This thesis investigates different approaches of improving the control performance, using nonlinear Model Predictive Control (MPC) as a foundation for the developed motion controllers due to its optimized solution and capability of satisfying constraints. First, a decoupled system is implemented and results are provided for two simple motion tasks showing problems related to the decoupling. Thereafter, two different approaches are taken to remedy the observed drawbacks. A nonlinear MPC controller is developed combining the motion controller and thruster allocation resulting in a more robust control system. Then, in order to keep the control system modularized, an investigation of possible ways to augment the decoupled system so as to achieve similar performance as the combined system is carried out. One proposed solution is a nonlinear MPC controller with time-varying constraints accounting for the current limitations of the thruster system. However, this did not always improve the control performance since the behavior of the TA still is unknown to the MPC controller.

Um dos desafios que ainda precisa ser superado com o objetivo de melhorar o desempenho do controle preditivo (MPC) é a sua manutenção. Reidentificação do processo é uma das melhores opções disponíveis para atualizar o modelo interno do MPC, a fim de melhorar seu desempenho. No entanto, o processo de reidentificação é dispendioso. Pesquisadores propuseram dois métodos diferentes, capazes de detectar divergências entre o processo real e o seu modelo, através da análise de correlações parciais. Utilizando essas técnicas, ao invés de reidentificar todos os sub-modelos do processo, apenas algumas entradas com divergência significativas teriam que ser perturbadas e somente a parte degradada do modelo seria atualizada. Entretanto, não há informações suficientes e análises sobre a influência das estruturas de modelo nos resultados das correlações parciais. Além disso, apesar de ambas as abordagens serem eficientes na detecção de divergências significativas, elas não fornecem informações suficientes sobre a sua quantificação. Esta dissertação de mestrado demonstra que o método de Carlsson (2010) é uma solução particular do método de Badwe et al. (2009), quando os modelos utilizados no processo de identificação são estruturas FIR. Além disso, alguns outros tipos de estruturas serão estudados, de modo a verificar se eles são adequados para a análise da correlação parcial, com o objetivo de detectar esse tipo de divergência. Quanto à limitação da detecção do nível da divergência entre o modelo e a planta, este trabalho propõe um projeto inicial de um novo método para resolver este problema, através da adição de ruído branco off-line nos dados coletados do processo, com diferentes variações antes da análise da correlação parcial. Um estudo de caso simulado é mostrado, que confirma a eficácia desta nova técnica. Finalmente, são apresentadas as conclusões encontradas e as possibilidades para estudos futuros. / One of the challenges that still needs to be overcome in order to improve the performance of the model predictive control (MPC) is its maintenance. Re-identification of the process is one of the best options available to update the internal model of the MPC, in order to improve performance. However, re-identification is costly. Researchers have proposed two different methods able to detect plant mismatch through partial correlation analysis. Using these techniques, instead of re-identifying all the sub-models in the process, only a few inputs with significant mismatch would have to be perturbed and only the degraded portion of the model would be updated. Nevertheless, there is not enough information and analysis about the influence of the model structures for identification on partial correlation results. Besides, although both approaches are efficient in detecting significant mismatches, they do not provide enough information about its magnitude. This masters thesis demonstrates that the Carlssons method (2010) is a particular solution of the Badwe et al.s method, when the models used on the identification process are FIR structures. Moreover, some other types of structures will be analyzed in order to check if they are suitable for the partial correlation procedure to detect plant mismatches. Concerning the limitation of the detection the level of plant-mismatch, this thesis proposes a starting project of a new method to address this issue by adding offline white noise to the collected data from the process with different variances before analyzing the partial correlation. A simulation case study is shown that confirms the efficacy of this new technique. Finally, conclusions and possible future studies are presented.

Neste trabalho foram desenvolvidas duas técnicas de sintonia para controladores preditivos por modelo. Ambas visam minimizar a soma do erro quadrático entre respostas do sistema em malha fechada e trajetórias de referência pré-definidas; a primeira resolve um problema de otimização lexicográfica enquanto a segunda resolve um problema de otimização de compromisso. As vantagens dos métodos apresentados são: maior automatização, definição de objetivos de sintonia intuitiva que considera especificações na dinâmica do processo, uma métrica no domínio do tempo e é capaz de incluir o conhecimento do engenheiro de controle em uma técnica de sintonia confiável. Um estudo de caso no sistema de craqueamento catalítico ilustrou a flexibilidade de definição dos objetivos da técnica lexicográfica. Um estudo de caso sobre uma coluna de fracionadora de óleo pesado em malha fechada com um controlador preditivo por modelo comparou ambas as estratégias de sintonia desenvolvidas aqui e pode-se concluir que a técnica lexicográfica dá prioridade aos objetivos importantes enquanto a técnica de compromisso calcula uma solução média, com respeito aos objetivos. A técnica de compromisso foi comparada a um método de sintonia da literatura quanto a aplicação em um controlador preditivo de horizonte infinito com targets para as entradas e controle por faixas das saídas com uma coluna de destilação. Observou-se que a técnica desenvolvida aqui é computacionalmente mais rápida e não requer a escolha de uma solução não-dominada dentre um conjunto de soluções de Pareto. Aplicações reais de controle preditivo são severamente afetadas por incerteza de modelo. Estendeu-se as técnicas desenvolvidas aqui para considerar o caso de incerteza multi-planta, calculando parâmetros de sintonia robustos para controladores nominais, visando tratar o compromisso entre performance e estabilidade e robustez da malha fechada. Um controlador preditivo de horizonte infinito foi sintonizado de forma robusta e comparado com um controlador preditivo robusto em malha fechada com um modelo de separadora C3/C4. Observou-se que este consegue controlar melhor o processo, entretanto, tem um tempo de computação duas ordens de grandeza maior que o controlador nominal, em operação on-line. / Two multi-objective optimization based tuning techniques for Model Predictive Control (MPC) were developed. Both take into account the sum of the squared errors between closed-loop trajectories and reference responses based on pre-defined goals as tuning objectives; one solves a lexicographic optimization to obtain an optimum set of tuning parameters (LTT), whereas the other solves a compromise optimization problem (CTT). The main advantages are an automated framework, and straightforward goal definition, which are capable of taking into account a specification on the process dynamics, a time-domain metrics, and of embedding the control engineers knowledge into a reliable approach. A fluid catalytic cracking tuning case study unveiled the goal definition flexibility of the LTT, with respect to output tracking and variable coupling. A heavy oil fractionator in closed-loop with a MPC case study compared both tuning techniques developed here, and it was observed that the LTT in fact prioritizes the main objectives, whereas the CTT yields an average solution, in terms of the tuning objectives. The CTT was compared to another multi-objective tuning technique from the literature, in the tuning of a MPC with input targets and output zone control in closed-loop with a crude distillation unit model. The simulation results showed that the CTT allows for faster results, regarding the computational time to compute the tuning parameters and there is no need of a posteriori decisions to select the best non-dominated solution. Real MPC applications are strongly hindered by model uncertainty. This limitation was addressed by the extension of the tuning techniques to account for multi-plant model uncertainty, thus obtaining optimum robustly tuned parameters for nominal controllers, addressing the trade-off between robustness and performance. A robustly tuned Infinite Horizon MPC (IHMPC) was compared to a Robust IHMPC, in closed-loop with a C3/C4 splitter system model. It was observed in a simulation that even though the latter yields better output responses, it is two orders of magnitude slower than the former in online operation.

Sistemas de pulverização com injeção direta possibilitam o uso de diferentes agrotóxicos em uma mesma aplicação, reduzindo o desperdício de agrotóxicos e minimizando desta forma os impactos toxicológico e ambiental relacionados com o preparo e descarte da calda. Neste trabalho foram desenvolvidos modelos matemáticos para um sistema de pulverização com injeção direta de agrotóxico, incluindo a dinâmica da concentração da calda. Também foi desenvolvida uma estratégia de controle preditivo com antecipação das taxas de aplicação para ajustar as taxas de aplicação do agrotóxico e da calda. Também, uma plataforma flexível para o desenvolvimento de pulverizadores foi projetada e construída. A sua automação foi baseada em um controlador embarcado de tempo real adequado para aplicações de controle, aquisição e temporização. Para obter os parâmetros dos modelos e avaliar a estratégia de controle ensaios de vazão e concentração para diferentes pontas de pulverização foram propostos. Com o emprego da abordagem de controle preditivo, os erros das vazões do agrotóxico e da calda ficaram abaixo do nível admissível de 5%. O uso da estratégia de antecipação das taxas de aplicação permitiu aumentar a eficiência da aplicação, reduzindo em até 40% os erros de aplicação. Resultados experimentais são apresentados para validar os modelos e mostrar a eficiência da estratégia de controle desenvolvida. / Sprayer systems with direct injection allow the use of different pesticides in a single application, reducing the waste of chemicals and thereby minimizing the toxicologic and environmental risks associated with the carrier-chemical mix preparation and discard. In this work, mathematical models for a direct chemical injection sprayer system including the dynamics of the carrier-chemical mix concentration are developed. Also, a predictive control strategy with anticipative reference of application rates was developed to adjust the carrier-chemical mix and chemical flow rates. Also, a flexible platform for the development of sprayers was designed and constructed. The automation of this platform was based on a programmable automation controller suitable for control, acquisition and timming applications. To obtain the models and analyse the control strategy, essays flow and concentration for different spray nozzles were proposed. With the use of predictive control approach, the errors of the carrier-chemical mix and chemical flow rates were lower than the admissible level of 5 %. The use of the advanced references increased the efficiency of the variable rate application, reducing up to 40 % application errors. Practical results are presented to validate the models and show the efficiency of the developed control strategy.

Este trabalho tem como objetivo principal o desenvolvimento de uma estratégia de integração da otimização com o controle preditivo multivariável em uma camada. Os problemas de controle e otimização econômica são resolvidos simultaneamente em um mesmo algoritmo. A função objetivo econômica foi inserida no controlador na sua forma diferencial, ou seja, o gradiente da função objetivo econômica. O método foi testado por simulação para o caso do sistema reator regenerador da UFCC (Unit of Fluid Catalytic Cracker). Esta dissertação descreve a estratégia de otimização integrada ao controlador preditivo cuja função objetivo incorpora componentes dinâmicos e estáticos. Para a determinação das condições ótimas do processo no estado estacionário do conversor (unidade de craqueamento catalítico) foi utilizado um modelo empírico do processo. A melhor trajetória para conduzir o processo para o seu ponto ótimo de operação, maximizando lucro ou produto de maior valor agregado, desde que não sejam violadas as restrições de processo, é predita utilizando um modelo dinâmico, obtido através de dados de testes em degrau em um modelo rigoroso. Este modelo linear possibilitou a obtenção das funções de transferência do processo e o modelo em variáveis de estado. O ponto ótimo que é obtido na execução deste algoritmo, leva em consideração a não violação das restrições das variáveis manipuladas e controladas do processo, tanto para o estado estacionário como para o transiente do problema. O problema de otimização não linear resultante é resolvido através de uma rotina de programação quadrática da biblioteca do Matlab. Uma segunda alternativa apresentada para a estratégia de otimização deste trabalho, é a inclusão do gradiente reduzido na função objetivo do controlador quando são observadas violações das restrições das variáveis controladas. Os resultados simulados através de um modelo não linear rigoroso (Moro&Odloak,1995) mostram um bom desempenho dos algoritmos aqui desenvolvidos tanto com relação aos benefícios econômicos como na estabilização da unidade. / This dissertation aims to develop a strategy to integrate the optimization problem of the plant into the model predictive controller in a one layer strategy, for the real time optimization or online optimization. The control and the optimization of the process are computed simultaneously in the same algorithm. The gradient of the economic objective function is included in the cost function of the controller instead of in its regular form. Thereby, this work describes a predictive control strategy, which can be classified as a one layer strategy and whose objective function has to be optimized obeying constraints, which incorporates dynamic and static components. The optimal conditions of the process in the steady state are defined through the use of an empirical process model. Furthermore, the best trajectory to be followed in order to reach the optimal conditions, without violating the constraints, maximizing profit or the production of its more valuable product, is predicted through the use of the dynamic model, that can be obtained through a plant step test. As a result transfer function and state space models are obtained. The optimal operation point is achieved through the execution of the proposed algorithm. Therefore, the solution to the optimization/control problem will always be in a feasible region, in other words, without violating the process manipulated or controlled variable constraints for both stationary and transient states of the problem. The non-linear optimization problem resulted from the implementation of the proposed algorithm is solved through the quadratic programming routine from the Matlab library. The second online optimization strategy proposed in this work is one that considers the reduced gradient method algorithm modified to evaluate the predicted trajectory. As a result, any violation of the manipulated or controlled variable constraints is prevented and this variable is not considered in the next step of the calculation of the predicted trajectory or even in the search direction of the optimization. Finally the simulations results obtained through the use of a nonlinear rigorous model (Moro&Odloak,1995) presents good performance for the algorithms here proposed, not only related to economic benefits, but also in order to stabilize the unit.

Uma das técnicas de controle avançado que vem ganhando destaque no cenário econômico e ecológico, focando maior sustentabilidade e a otimização dos processos, é o controle preditivo, o qual já vem sendo aplicado em indústrias químicas e petroquímicas. Esta dissertação trata do desenvolvimento de um controle preditivo aplicado a uma planta piloto de neutralização de pH, presente no Laboratório de Controle de Processos Industriais da Escola Politécnica da Universidade de São Paulo. O desenvolvimento do projeto pode ser dividido em quatro etapas: implementação das malhas de controle regulatório, identificação dos sistemas, construção do controlador preditivo, aplicações e análises experimentais. Na primeira etapa foi necessário estudar o sistema em questão e implementar algumas malhas internas usando controladores PID. Na segunda etapa foi realizada a identificação do modelo da planta, ressaltando que pontos de operação e ajuste de parâmetros internos são determinantes para a modelagem. Já na terceira etapa desenvolveu-se um controlador preditivo, através de softwares auxiliares como o MATLAB e o IIT 800xA da ABB, que foram utilizados para o desenvolvimento e implementação do algoritmo de controle. Por fim, na última etapa, foi feita a análise e comparação dos resultados, quando se submete à planta a um controlador PID, quando aplicado um controlador preditivo em cascata com controladores PID e quando se utiliza apenas o controlador preditivo com ação direta nos atuadores. / The predictive control is an advanced control technique which has gained evidence in the economic and ecological context because the search for sustainability and process optimization. This control has already been applied by the chemical and petrochemical industries. The purpose of this project is to develop a predictive controller which will be applied in a pH neutralization plant located in the Industrial Processes Control Laboratory at Polytechnic School of the University of São Paulo. The development of this project can be divided into four stages: implementation of regulatory control loops, identification of the system, construction of the predictive controller, applications and experimental analysis. The first step is necessary in order to study the plant and to implement some internal loops using PID controllers. In the second step, the identification process of the plant model will be done. It is important to note that operating points and internal parameter settings are very important for modeling. In the third stage, using the model obtained from the identification process, a predictive controller is built from auxiliary software such as MATLAB and IIT 800xA (by ABB), which will be used for the development and implementation of the control algorithm. Finally, the last step consists in collecting and analyzing the results of the pH neutralization plant. At this stage the responses of each controller will be compared: PID controller, MPC controller in cascade mode with PID and MPC controller acting directly on actuators.