The freight transport plays an important role in the development of the economy. However, this comes with an important contribution to greenhouse gas emission. Recently a shift toward heavy-duty electric vehicles has been made, but some issues still need to be tackled. One of them is to develop ways to quickly recharge the vehicle’s batteries without damaging them. In this thesis, we highlight that not only the current but also the battery temperature need to be carefully managed in order to prevent damages during a charging session. To show that, an electrical, thermal and aging model of Li-ion battery is developed. A charging strategy based on a Model Predictive Control algorithm is proposed. The algorithm controls both the battery current and cooling system in order to achieve the optimal balance between the charging speed and the preservation of the battery. The resulting algorithm is tested, in simulation, against a conventional constant current charging in different charging scenarios. The results show an important increase in performance and highlight the role of the battery cooling system in the preservation of the battery. / Godstransporterna spelar en viktig roll för ekonomins utveckling. Detta innebär dock ett betydande bidrag till utsläppen av växthusgaser. På senare tid har en övergång till tunga elfordon skett, men vissa frågor måste fortfarande lösas. En av dem är att utveckla metoder för att snabbt ladda fordonsbatterierna utan att skada dem. I den här avhandlingen lyfter vi fram att inte bara strömmen utan även batteritemperaturen måste hanteras noggrant för att förhindra skador under en laddning. För att visa detta utvecklas en elektrisk, termisk och åldrande modell för Li-ion-batterier. En laddningsstrategi baserad på en algoritm för modellförutsägbar styrning föreslås. Algoritmen styr både batteriströmmen och kylsystemet för att uppnå en optimal balans mellan laddningshastighet och bevarande av batteriet. Den resulterande algoritmen testas i simulering mot en konventionell konstantström laddning i olika laddningsscenarier. Resultaten visar en betydande ökning av prestanda och belyser batterikylsystemets betydelse för bevarandet av batteriet.

Temperature control is a widely researched topic and a common application is in heating systems such as buildings. A temperature control method that is central in ensuring comfort and reduction of energy consumption in modern buildings and other heating systems is based on model predictive control (MPC). Traditionally, the MPC optimal control problem is to track a target, but there are other examples of optimization problems besides tracking problems and one such optimization problem is the economical optimization problem, an optimization based on economical objectives. A heating system with electrical supply may be controlled by an economic MPC such that the economical objective is to consider time-varying prices of electricity.  This thesis studies how time-varying prices of electricity can be utilized as an economical objective in an economical MPC to reduce electricity costs for a double-heater system. This is done using an available model of the double-heater system and an MPC to construct an economical MPC. The performance of the economical MPC is then investigated and compared to the existing MPC.  In the thesis it is found, through a test with six different cost profiles and a test with historical data of forecasts of electricity prices, that the economical MPC can reduce total electricity costs when compared to the existing MPC. Furthermore it is found that the performance of the economic MPC is acceptable when it is compared with and without prediction of setpoint changes, prediction of price changes and an isolating layer between the heaters. The thesis concludes that satisfactory results are attained, as the economical MPC leads to decreased total electricity costs for the double-heater system and notes that the economic MPC is versatile by accepting both user-defined and historical cost profiles.

<p dir="ltr">Supervisory predictive control of residential building heating, ventilation, and air conditioning (HVAC) systems could protect electrical infrastructure, enhance occupants’ thermal comfort, reduce energy costs, and minimize emissions. However, there are few experimental demonstrations, with most of the work focusing on simulation studies. To convince stakeholders of the benefits of supervisory predictive controls for residential HVAC systems, it is important to demonstrate practical systems in real buildings. Practical demonstrations also further our understanding of the field performance of these systems. This thesis presents the first comprehensive review of supervisory predictive control experiments in residential buildings, drawing critical insights on the estimated energy savings, the types of equipment controlled, the objectives and problem formulations considered, and other practical considerations. To address limitations in the existing body of experimental work, a series of field demonstrations were performed in a real house with student occupants near the Purdue campus in West Lafayette, Indiana, U.S.A.</p><p dir="ltr">The first field demonstration involved supervisory predictive control of an air-to-air heat pump with backup electric resistance heat. This was the first experiment to consider this equipment configuration, which is common in North America. A simple data-driven method is presented for learning a model of the temperature dynamics of a detached residential building. Using this model, the control system adjusts indoor temperature set points based on weather forecasts, occupancy conditions, and data-driven models of the heating equipment. Field tests from January to March of 2023 included outdoor temperatures as low as −15 ℃. During these tests, the control system reduced total heating energy costs by 19% on average (95% confidence interval: 13–24%) and energy used for backup heat by 38%. The control system also reduced the frequency of using high-stage (19 kW) backup heat by 83%. Concurrent surveys of residents showed that the control system maintained satisfactory thermal comfort. These real-world results could strengthen the case for deploying predictive home heating control, bringing the technology one step closer to reducing emissions, utility bills, and power grid impacts at scale.</p><p dir="ltr">The second field demonstration advanced the state of the art of predictive residential cooling control, wherein past experimental demonstrations relied on “sensible” models of building thermal dynamics and neglected humidity effects. In this thesis, a model-free machine learning method is introduced to predict the indoor wet-bulb temperature and the sensible heat ratio in a “latent” model formulation, with the aim to increase the accuracy of the real electrical power prediction. The latent and sensible formulations are tested in two separate model predictive controller (MPC) schemes in an on-off fashion. One MPCscheme aims to reduce energy costs while enhancing comfort. The other is a power-limiting controller that aims to keep the power of the HVAC equipment below 2.5 kW between 4 PM and 8 PM. The two MPC schemes and the two load models are assessed through 38 days of testing. It is found that across both economic MPC and power-limiting MPC, the energy savings across the latent and sensible formulations are similar. Through a normalized Cooling Degrees Days analysis, the energy savings to the baseline controller in the house are found to be 16 to 32% for economic MPC (95% confidence interval) and -5 to 10% for power-limiting MPC, with 7 to 21% savings across both controllers (14% mean). For power limiting, the latent formulation reduced the total duration of constraint violation by 88% and the sensible formulation by 40%, with respect to the non-MPC baseline. Additionally, the latent formulation reduced the peak power demand by 13% relative to the baseline, a behavior not observed in the sensible formulation.</p><p dir="ltr">The third field experiment investigated the problem of protecting home electrical infrastructure in the context of electrification retrofits. Installing electric appliances or vehicle charging in a residential building can sharply increase the electric current draws. In older housing, high current draws can jeopardize circuit breaker panels or electrical service (the wires that connect a building to the distribution grid). Upgrading electrical panels or service often entails long delays and high costs, and thus it poses a significant barrier to electrification. This thesis develops and field tests a novel control system that avoids the need for electrical upgrades by maintaining an electrified home’s total current draw within the safe limits of its existing panel and service. In the proposed control architecture, a high-level controller plans device set-points over a rolling prediction horizon, while a low-level controller monitors real-time conditions and ramps down devices if necessary. The control system was tested for 31 consecutive winter days with outdoor temperatures as low as -20 ℃. The control system maintained the whole-home current within the safe limits of electrical panels and service rated at 100 A, a common rating for older houses in North America, by adjusting only the temperature set-points of the heat pump and water heater. Simulations suggest that the same 100 A limit could accommodate a second electric vehicle (EV) with Level II (11.5 kW) charging. The proposed control system could allow older homes to safely electrify without upgrading electrical panels or service, saving a typical household on the order of $2,000 to $10,000. </p><p dir="ltr">These three field experiments demonstrate that low-cost predictive control systems can serve multiple objectives, improving the efficiency of heat pumps and water heaters while maintaining comfort and protecting electrical infrastructure. Future work will be directed toward improving the scalability of these proposed controllers through the incorporation of data-driven methodologies such as data-enabled predictive control, as well as understanding the application of these algorithms with different systems, including batteries, on-site solar photovoltaics, and electrical vehicle charging.</p>

The aim of this project was to compare two control algorithms designed to steer an autonomous vehicle. The comparison was made using a simulated environment to evaluate the performance of both controllers. The simulation used in this project was designed in Python and used an algorithm which randomly constructed roads from predefined road segments to create paths for the vehicle to follow. In this environment the Linear Time Varying (LTV)-Model Predictive Controller (MPC) and Pure Pursuit Controller (PPC) algorithms were evaluated. The thesis compared how well they follow paths, the average control cost of completing tasks, how well they handle input constraints, and the computational time for each algorithm. The data was collected by driving along three sets of randomly generated roads with both control algorithms. One set mostly straight, one with some turns, and one with mostly turns. An Analysis of Variance (ANOVA) test was used to make the comparison between the performance of the two algorithms. The results showed that both algorithms performed well. The PPC had low computation time and used less control, but it also had larger position errors. The LTV-MPC had higher computation time, but smaller position errors at the cost of larger control values. The conclusion is that the MPC is preferable if computational capabilities are available. Room for future work exists in the form of comparing additional controller types for autonomous vehicles and exploring different tuning parameters for the MPC controller. The simulation could also be expanded to more accurately reflect real world conditions. / Målet med detta projekt var att jämföra två kontrollalgoritmer avsedda för att styra en självkörande bil. Jämförelsen gjordes med hjälp av en simulering som utformades i Python. Den använde sig av en algoritm som slumpmässigt satte ihop vägar från förkonstruerade delar för att skapa banor för den självkörande bilen att följa. I denna miljö har vi testat två algoritmer, en LTV-MPC och en PPC. Vi jämförde hur pass väl de följer banor som skall likna riktiga vägar, hur mycket styrning de använder sig av för att bedöma energianvändning, hur väl de förhåller sig till begränsningar på acceleration och styrning, och den beräkningstiden som krävdes för att köra vår algoritm. Datan samlades genom att köra längs med tre grupper av slumpmässigt genererade vägar med båda kontrollalgoritmerna. En grupp innehöll huvudsakligen raka sträckor, en innehöll en del svängar, och en innehöll mycket svängar. ANOVA-testet användes för att göra jämförelsen mellan resultatet av dessa två algoritmer. Resultatet visade att båda algoritmer presterar väl. PPCn hade låg beräkningstid och mindre styrvärden, men större positionsfel. MPCn hade högre beräkningstid och större styrvärden, men mindre positionsfel. Slutsatsen är att MPCn är att föredra om beräkningsmöjligheterna finns tillgängliga. Det finns utrymme för framtida arbete i form av att jämföra fler kontrollalgoritmer och att utforska fler parameter justeringar för MPCn. Utöver det finns det även utrymme för en simulation som reflekterar verkligheten noggrannare.

Cette thèse vise à élaborer de nouvelles stratégies de commande basées sur la commande prédictive pour le système de génération d’énergie éolienne. La topologie des systèmes de production éolienne basées sur le Générateur Asynchrone à Double Alimentation (GADA) qui convient à des plateformes de génération dans la gamme de puissance de 1.5 à 6 MW est abordée. Du point de vue technologique, le convertisseur à trois niveaux et clampé par le neutre (3L-NPC) est considéré comme une bonne solution pour une puissance élevée en raison de ses avantages: capacité à réduire la distorsion harmonique de la tension de sortie et du courant, et augmentation de la capacité du convertisseur grâce à une tension réduite appliquée à chaque semi-conducteur de puissance. Une description détaillée de la commande prédictive à ensemble de commande fini (FCS-MPC) avec un horizon de prédiction de deux pas est présentée pour deux boucles de régulation: celle liée au convertisseur connecté au réseau et celle du convertisseur connecté au GADA. Le principe de la commande repose sur l’utilisation d’un modèle de prédiction permettant de prédire le comportement du système pour chaque état de commutation du convertisseur. La minimisation d’une fonction de coût appropriée prédéfinie permet d’obtenir la commutation optimale à appliquer au convertisseur. La thèse étudie premièrement les problèmes liées à la compensation du temps de calcul de la commande et au choix et aux pondérations de la fonction de coût. Ensuite, le problème de stabilité de la commande FCS-MPC est abordé en considérant une fonction de Lyapunov dans la minimisation de la fonction de coût. Finalement, une étude sur la compensation des effets des temps morts du convertisseur est présentée. / This thesis aims to elaborate new control strategies based on Model Predictive control for wind energy generation system. We addressed the topology of doubly fed induction generator (DFIG) based wind generation systems which is suitable for generation platform power in the range in 1.5-6 MW. Furthermore, from the technological point of view, the three-level neutral-point clamped (3L-NPC) inverter configuration is considered a good solution for high power due to its advantages: capability to reduce the harmonic distortion of the output voltage and current, and increase the capacity of the converter thanks to a decreased voltage applied to each power semiconductor.In this thesis, we presented a detailed description of finite control set model predictive control (FCS-MPC) with two step horizon for two control schemes: grid and DFIG connected 3L-NPC inverter. The principle of the proposed control scheme is to use system model to predict the behaviour of the system for every switching states of the inverter. Then, the optimal switching state that minimizes an appropriate predefined cost function is selected and applied directly to the inverter.The study of issues such as delay compensation, computational burden and selection of weighting factor are also addressed in this thesis. In addition, the stability problem of FCS-MPC is solved by considering the control Lyapunov function in the design procedure. The latter study is focused on the compensation of dead-time effect of power converter.

In recent years, mobile robots have gained tremendous attention from the entire society: the industry is aiming at selling more intelligent products while the academia is improving their performance from all perspectives. Real world examples include autnomous driving vehicles, multirotors, legged robots, etc. One of the challenging tasks commonly faced by all game players, and all robotics platforms, is to plan motion or locomotion of the robot, calculate an optimal trajectory according to certain criterion and control it accordingly. Difficulty of solving such task usually arises from high-dimensionality and complexity of the system dynamics, fast changing conditions imposed as constraints and necessity for real-time deployment. This work proposes a method over the aforementioned mission by solving an optimal control problem in a receding horizon fashion. Unlike the existing Sequential Linear Quadratic [1] algorithm which is a continuous-time variant of Differential Dynamic Programming [2], we tackle the problem in a discretized multiple shooting fashion. Sequential Quadratic Programming is employed as optimization technique to solve the constrained Nonlinear Programming iteratively. Moreover, we apply trust region method in the sub Quadratic Programming to handle potential indefiniteness of Hessian matrix as well as to improve robustness of the solver. Simulation and benchmark with previous method have been conducted on robotics platforms to show the effectiveness of our solution and superiority under certain circumstances. Experiments have demonstrated that our method is capable of generating trajectories under complicated scenarios where the Hessian matrix contains negative eigenvalues (e.g. obstacle avoidance). / De senaste åren har mobila robotar fått enorm uppmärksamhet från hela samhället: branschen siktar på att sälja mer intelligenta produkter samtidigt som akademin förbättrar sina prestationer ur alla perspektiv. Exempel på verkligheten inkluderar autonoma körande fordon, multirotorer, robotar med ben, etc. En av de utmanande uppgifterna som vanligtvis alla spelare och alla robotplattformar står inför är att planera robotens rörelse eller rörelse, beräkna en optimal bana enligt vissa kriterier och kontrollera det därefter. Svårigheter att lösa en sådan uppgift beror vanligtvis på hög dimensionalitet och komplexitet hos systemdynamiken, snabbt föränderliga villkor som åläggs som begränsningar och nödvändighet för realtidsdistribution. Detta arbete föreslår en metod över det tidigare nämnda uppdraget genom att lösa ett optimalt kontrollproblem på ett vikande horisont. Till skillnad från den befintliga Sequential Linear Quadratic [1] algoritmen som är en kontinuerlig tidsvariant av Differential Dynamic Programming [2], tar vi oss an problemet på ett diskretiserat multipelfotograferingssätt. Sekventiell kvadratisk programmering används som optimeringsteknik för att lösa den begränsade olinjära programmeringen iterativt. Dessutom tillämpar vi trust region-metoden i den sub-kvadratiska programmeringen för att hantera potentiell obestämdhet av hessisk matris samt för att förbättra lösarens robusthet. Simulering och benchmark med tidigare metod har utförts på robotplattformar för att visa effektiviteten hos vår lösning och överlägsenhet under vissa omständigheter. Experiment har visat att vår metod är kapabel att generera banor under komplicerade scenarier där den hessiska matrisen innehåller negativa egenvärden (t.ex. undvikande av hinder).

Tesis por compendio / [ES] Esta tesis está dedicada al desarrollo de conceptos de control predictivo basados en modelos para la compensación de armónicos en sistemas de potencia con fuentes de energía renovables. En concreto, estos conceptos proporcionan una corriente de compensación de referencia para un filtro activo de potencia conectado en el punto de acoplamiento común, mejorando así la calidad de potencia del sistema. No obstante, los resultados podrían aplicarse en general a problemas de control en los que el objetivo sea seguir la forma de una determinada señal. La tesis propone dos métodos principales de control basados en la teoría de control predictivo de modelos (MPC). El primer controlador, es decir, el control predictivo de modelos con conformación de señal de estado lineal (LS3MPC), se basa en la teoría MPC cuadrática estándar. Sin embargo, contrariamente a la práctica habitual de control de referencia fija, el LS3MPC incorpora la dinámica deseada del sistema directamente en su función de coste, utilizando los denominados residuos de clase de forma lineal. Este enfoque permite que la función de costes del LS3MPC sea más adaptable y ofrezca más compensaciones dinámicas, especialmente cuando está sometida a restricciones. Al utilizar los residuos de clase de forma, el problema MPC garantiza que la planta controlada siga la dinámica dada por la clase de forma. En este caso, la dinámica deseada está determinada por la clase de forma armónica lineal propuesta, es decir, la dinámica de una señal armónica fundamental de frecuencia fija. Desde el punto de vista de la implementación, se propone una formulación MPC explícita para el LS3MPC con el fin de mejorar su aplicabilidad en tiempo real. El LS3MPC explícito propuesto utiliza un enfoque de malla equidistante en formato tensorial para aproximar la solución MPC explícita con afinidad por partes. Usando la descomposición tensorial, el LS3MPC explícito puede romper la maldición de la dimensionalidad, reduciendo significativamente la carga de memoria y trivializando el problema en tiempo real de localización de puntos. El segundo controlador, es decir, el control predictivo de modelo de ciclo límite (LCMPC), se centra en resolver las deficiencias del LS3MPC. En concreto, el LCMPC aborda la falta de control directo de la amplitud recurriendo a la teoría MPC no lineal. El LCMPC introduce una clase de forma armónica no lineal basada en una forma normal de bifurcación supercrítica de Neimark-Sacker. Al igual que el LS3MPC, el LCMPC también incorpora el residual de su clase de forma armónica no lineal directamente en su función de coste, proporcionando las mismas ventajas mencionadas anteriormente. En cuanto a la estabilidad del sistema, se desarrollan condiciones suficientes, para un estado inicial predeterminado, que garanticen que el sistema de bucle cerrado permanece dentro de la región de atracción de la forma normal ante una perturbación suficientemente pequeña. Ambos controladores se someten a pruebas con estudios de simulación en múltiples escenarios, proporcionando resultados de compensación consistentemente satisfactorios. / [CA] Aquesta tesi està dedicada al desenvolupament de conceptes de control predictiu basats en models per a la compensació d'harmònics en sistemes de potència amb fonts d'energia renovables. En concret, aquests conceptes proporcionen un corrent de compensació de referència per a un filtre actiu de potència connectat en el punt d'acoblament comú, millorant així la qualitat de potència del sistema. No obstant això, els resultats podrien aplicar-se en general a problemes de control en els quals l'objectiu siga seguir la forma d'un determinat senyal. La tesi proposa dos mètodes principals de control basats en la teoria de control predictiu de models (MPC). El primer controlador, és a dir, el control predictiu de models amb conformació de senyal d'estat lineal (LS3MPC), es basa en la teoria MPC quadràtica estàndard. No obstant això, contràriament a la pràctica habitual de control de referència fixa, el LS3MPC incorpora la dinàmica desitjada del sistema directament en la seua funció de cost, utilitzant els denominats residus de classe de manera lineal. Aquest enfocament permet que la funció de costos del LS3MPC siga més adaptable i oferisca més compensacions dinàmiques, especialment quan està sotmesa a restriccions. En utilitzar els residus de classe de forma, el problema MPC garanteix que la planta controlada seguisca la dinàmica donada per la classe de forma. En aquest cas, la dinàmica desitjada està determinada per la classe de manera harmònica lineal proposta, és a dir, la dinàmica d'un senyal harmònic fonamental de freqüència fixa. Des del punt de vista de la implementació, es proposa una formulació MPC explícita per al LS3MPC amb la finalitat de millorar la seua aplicabilitat en temps real. El LS3MPC explícit proposat utilitza un enfocament de malla equidistant en format tensorial per a aproximar la solució MPC explícita amb afinitat per parts. Usant la descomposició tensorial, el LS3MPC explícit pot trencar la maledicció de la dimensionalitat, reduint significativament la càrrega de memòria i trivialitzant el problema en temps real de localització de punts. El segon controlador, és a dir, el control predictiu de model de cicle límit (LCMPC), se centra en resoldre les deficiències del LS3MPC. En concret, el LCMPC aborda la falta de control directe de l'amplitud recorrent a la teoria MPC no lineal. El LCMPC introdueix una classe de manera harmònica no lineal basada en una forma normal de bifurcació supercrítica de Neimark-Sacker. Igual que el LS3MPC, el LCMPC també incorpora el residual de la seua classe de manera harmònica no lineal directament en la seua funció de cost, proporcionant els mateixos avantatges esmentats anteriorment. Quant a l'estabilitat del sistema, es desenvolupen condicions suficients, per a un estat inicial predeterminat, que garantisquen que el sistema de bucle tancat roman dins de la regió d'atracció de la forma normal davant una pertorbació prou xicoteta. Tots dos controladors se sotmeten a proves amb estudis de simulació en múltiples escenaris, proporcionant resultats de compensació consistentment satisfactoris. / [EN] This thesis is dedicated to developing model-based predictive control concepts for harmonic compensation in power systems with renewable energy sources. Specifically, these concepts provide a reference compensation current for an active power filter connected at the point of common coupling, thereby enhancing the system's power quality. Nevertheless, results could be generically applied to control problems where the task is to follow a certain shape of a signal. The thesis proposes two main control approaches based on model predicitve control (MPC) theory. The first controller, i.e., the linear state signal shaping model predictive control (LS3MPC), relies on standard quadratic MPC theory. However, contrary to standard fixed reference control practice, the LS3MPC embeds the desired system dynamics directly into its cost function, using the so-called linear shape class residuals. This approach allows the LS3MPC' cost function to be more adaptive, providing more dynamic trade-offs, especially when constrained. By using shape class residuals, the MPC problem ensures that the controlled plant follows the desired dynamics given by the shape class. In this case, the target dynamics are given by the proposed linear harmonic shape class, i.e, the dynamics of a fundamental harmonic signal of fixed frequency. From an application perspective, an explicit MPC formulation for the LS3MPC is proposed to enhance its real-time applicability. The proposed explicit LS3MPC uses an equidistant mesh grid approach in tensor format to approximate the piecewise affine explicit MPC solution. Using tensor decomposition, the explicitLS3MPC can break the curse of dimensionality, significantly reducing memory burden and trivializing the online point localization problem. The second controller, i.e., the limit cycle model predictive control (LCMPC), focuses on addressing the shortcomings of the LS3MPC. Namely, the LCMPC addresses the lack of direct amplitude control by reaching into nonlinear MPC theory. The LCMPC introduces a nonlinear harmonic shape class based on a supercritical Neimark-Sacker bifurcation normal form. Similarly to theLS3MPC, the LCMPC also embeds its nonlinear harmonic shape class residual directly in its cost function, providing the same benefits mentioned before. Regarding system stability, sufficient conditions are developed for a given initial state to ensure that the closed-loop system remains inside the normal form region of attraction for a sufficiently small disturbance. Both controllers are tested with simulation studies in multiple scenarios, providing consistently satisfactory compensation results. / The contributions on this doctoral thesis were partly developed at the Fraunhofer Institute for Silicon Technology ISIT within the project North German Energy Transition 4.0 (ger- man: Norddeutsche EnergieWende, NEW 4.0), which is funded by the German Federal Ministry for Economic Affairs and Energy (german: Bundesministerium für Wirtschaft und Energie, BMWi. This work was also partly funded with the project Northern German Living Lab (german: Norddeutsches Reallabor, NRL) by the Federal Ministry for Economic Affairs and Climate Action, by Generalitat Valenciana regional government through project CIAICO/2021/064, and by the Free and Hanseatic City of Hamburg (Hamburg City Parliament publication 20/11568). / Cateriano Yáñez, C. (2024). State Shaping Model Predictive Control for Harmonic Compensation [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/209406 / Compendio

The developments of advanced non-linear control strategies have attracted a considerable research interests over the past decades especially in process control. Rather than an absolute reliance on mathematical models of process plants which often brings discrepancies especially owing to design errors and equipment degradation, non-linear models are however required because they provide improved prediction capabilities but they are very difficult to derive. In addition, the derivation of the global optimal solution gets more difficult especially when multivariable and non-linear systems are involved. Hence, this research investigates soft computing techniques for the implementation of a novel real time constrained non-linear model predictive controller (NMPC). The time-frequency localisation characteristics of wavelet neural network (WNN) were utilised for the non-linear models design using system identification approach from experimental data and improve upon the conventional artificial neural network (ANN) which is prone to low convergence rate and the difficulties in locating the global minimum point during training process. Salient features of particle swarm optimisation and a genetic algorithm (GA) were combined to optimise the network weights. Real time optimisation occurring at every sampling instant is achieved using a GA to deliver results both in simulations and real time implementation on coupled tank systems with further extension to a complex quadruple tank process in simulations. The results show the superiority of the novel WNN-NMPC approach in terms of the average controller energy and mean squared error over the conventional ANN-NMPC strategies and PID control strategy for both SISO and MIMO systems.

Les exigences croissantes sur l'efficacité énergétique des bâtiments, l'évolution du {marché} énergétique, le développement technique récent ainsi que les particularités du poste de chauffage ont fait du MPC le meilleur candidat pour la régulation thermique des bâtiments à occupation intermittente. Cette thèse présente une méthodologie basée sur la commande prédictive distribuée visant un compromis entre l'optimalité, la simplicité et la flexibilité de l'implantation de la solution proposée. Le développement de l'approche est progressif : à partir du cas d'une seule zone, la démarche est ensuite étendue au cas multizone et / ou multisource, avec la prise en compte des couplages thermiques entre les zones adjacentes. Après une formulation quadratique du critère MPC pour mieux satisfaire les objectifs économiques du contrôle, la formulation linéaire est retenue. Pour répartir la charge de calcul, des méthodes de décomposition linéaire (comme Dantzig-Wolfe et Benders) sont employées. L'efficacité des algorithmes distribués proposés est illustrée par diverses simulations. / The increasing requirements on energy efficiency of buildings, the evolution of the energy market, the technical developments and the characteristics of the heating systems made of MPC the best candidate for thermal control of intermittently occupied buildings. This thesis presents a methodology based on distributed model predictive control, aiming a compromise between optimality, on the one hand, and simplicity and flexibility of the implementation of the proposed solution, on the other hand. The development of the approach is gradually. The mono-zone case is initially considered, then the basic ideas of the solution are extended to the multi-zone and / or multi-source case, including the thermal coupling between adjacent zones. Firstly we consider the quadratic formulation of the MPC cost function, then we pass towards a linear criterion, in order to better satisfy the economic control objectives. Thus, linear decomposition methods (such as Dantzig-Wolfe and Benders) represent the mathematical tools used to distribute the computational charge among the local controllers. The efficiency of the distributed algorithms is illustrated by simulations.

Cette thèse porte sur l’application des approches de commande prédictive pour l’optimisation des démarrages des centrales à cycles combinés. Il s’agit d’une problématique à fort enjeu qui pose des défis importants. L’élaboration des approches est progressive. Dans une première partie un modèle de centrale est construit et adapté à l’optimisation, en utilisant une méthodologie qui transforme des modèles physiques Modelica conçus pour la simulation en des modèles pour l’optimisation. Cette méthodologie a permis de construire une bibliothèque adaptée à l’optimisation. La suite des travaux porte sur l’utilisation du modèle afin d’optimiser phase par phase les performances du démarrage. La solution proposée optimise, en temps continu, le profil de charge des turbines en recherchant dans des ensembles de fonctions particulières. Le profil optimal est déterminé en considérant que celui-ci peut être décrit par une fonction paramétrée dont les paramètres sont calculés en résolvant un problème de commande optimale sous contraintes. La dernière partie des travaux consiste à intégrer cette démarche d’optimisation à temps continu dans une stratégie de commande à horizon glissant. Cette approche permet d’une part de corriger les dérives liées aux erreurs de modèles et aux perturbations, et d’autre part, d’améliorer le compromis entre le temps de calcul et l’optimalité de la solution. Cette approche de commande conduit cependant à des temps de calcul importants. Afin de réduire le temps de calcul, une structure de commande prédictive hiérarchisée avec deux niveaux, en travaillant à des échelles de temps et sur des horizons différents, a été proposée. / This thesis focuses on the application of model predictive control approaches to optimize the combined cycle power plants start-ups. Generally, the optimization of start-up is a very problematic issue that poses significant challenges. The development of the proposed approaches is progressive. In the first part a physical model of plant is developed and adapted to optimization purposes, by using a methodology which transforms Modelica model components into optimization-oriented models. By applying this methodology, a library suitable for optimization purposes has been built.In the second part, based on the developed model, an optimization procedure to improve the performances of the start-up phases is suggested. The proposed solution optimizes, in continuous time, the load profile of the turbines, by seeking in specific sets of functions. The optimal profile is derived by considering that this profile can be described by a parameterized function whose parameters are computed by solving a constrained optimal control problem. In the last part, the open-loop optimization procedure has been integrated into a receding horizon control strategy. This strategy represents a robust solution against perturbation and models errors, and enables to improve the trade-off between computation time and optimality of the solution. Nevertheless, the control approach leads to a significant computation time. In order to obtain real-time implementable results, a hierarchical model predictive control structure with two layers, working at different time scales and over different prediction horizons, has been proposed.