abstract: There is increasing interest in the medical and behavioral health communities towards developing effective strategies for the treatment of chronic diseases. Among these lie adaptive interventions, which consider adjusting treatment dosages over time based on participant response. Control engineering offers a broad-based solution framework for optimizing the effectiveness of such interventions. In this thesis, an approach is proposed to develop dynamical models and subsequently, hybrid model predictive control schemes for assigning optimal dosages of naltrexone, an opioid antagonist, as treatment for a chronic pain condition known as fibromyalgia. System identification techniques are employed to model the dynamics from the daily diary reports completed by participants of a blind naltrexone intervention trial. These self-reports include assessments of outcomes of interest (e.g., general pain symptoms, sleep quality) and additional external variables (disturbances) that affect these outcomes (e.g., stress, anxiety, and mood). Using prediction-error methods, a multi-input model describing the effect of drug, placebo and other disturbances on outcomes of interest is developed. This discrete time model is approximated by a continuous second order model with zero, which was found to be adequate to capture the dynamics of this intervention. Data from 40 participants in two clinical trials were analyzed and participants were classified as responders and non-responders based on the models obtained from system identification. The dynamical models can be used by a model predictive controller for automated dosage selection of naltrexone using feedback/feedforward control actions in the presence of external disturbances. The clinical requirement for categorical (i.e., discrete-valued) drug dosage levels creates a need for hybrid model predictive control (HMPC). The controller features a multiple degree-of-freedom formulation that enables the user to adjust the speed of setpoint tracking, measured disturbance rejection and unmeasured disturbance rejection independently in the closed loop system. The nominal and robust performance of the proposed control scheme is examined via simulation using system identification models from a representative participant in the naltrexone intervention trial. The controller evaluation described in this thesis gives credibility to the promise and applicability of control engineering principles for optimizing adaptive interventions. / Dissertation/Thesis / M.S. Electrical Engineering 2011

Cette thèse aborde les problèmes de la modélisation et de la commande d'un micro-réseau courant continu (CC) en vue de la gestion énergétique optimale, sous contraintes et incertitudes. Le micro-réseau étudie contient des dispositifs de stockage électrique (batteries ou super-capacités), des sources renouvelables (panneaux photovoltaïques) et des charges (un système d'ascenseur motorise par une machine synchrone a aimant permanent réversible). Ces composants, ainsi que le réseau triphasé, sont relies a un bus commun en courant continu, par des convertisseurs dédies. Le problème de gestion énergétique est formule comme un problème de commande optimale qui prend en compte la dynamique du système, des contraintes sur les variables, des prédictions sur les prix, la consommation ou la production et des profils de référence.Le micro-réseau considère est un système complexe, de par l'hétérogénéité de ses composants, sa nature distribuée, la non-linéarité de certaines dynamiques, son caractère multi-physiques (électromécanique, électrochimique, électromagnétique), ainsi que la présence de contraintes et d'incertitudes. La représentation consistante des puissances échangées et des énergies stockées, dissipées ou fournies au sein de ce système est nécessaire pour assurer son opération optimale et fiable.Le problème pose est abordé via l'usage combine de la formulation hamiltonienne a port, de la platitude et de la commande prédictive économique base sur le modelé. Le formalisme hamiltonien a port permet de décrire les conservations de la puissance et de l'énergie au sein du micro-réseau explicitement et de relier les composants hétérogènes dans un même cadre théorique. Les non linéarités sont gérées par l'introduction de la notion de platitude démentielle et la sélection de sorties plates associées au modèle hamiltonien a ports. Les profils de référence sont génères a l'aide d'une para métrisation des sorties plates de telle sorte que l'énergie dissipée soit minimisée et les contraintes physiques satisfaites. Les systèmes hamiltoniens sur graphes sont ensuite introduits pour permettre la formulation et la résolution du problème de commande prédictive _économique a l'échelle de l'ensemble du micro-réseau CC. Les stratégies de commande proposées sont validées par des résultats de simulation pour un système d'ascenseur multi-sources utilisant des données réelles, identifiées sur base de mesures effectuées sur une machine synchrone. / The goals of this thesis is to propose modelling and control solutions for the optimal energy management of a DC microgrid under constraints. The studied microgrid system includes electrical storage units (e.g., batteries, supercapacitors), renewable sources (e.g., solar panels) and loads (e.g., an electro-mechanical elevator system). These interconnected components are linked to a three phase electrical grid through a DC bus and associated DC/AC converters. The optimal energy management is usually formulated as an optimal control problem which takes into account the system dynamics, cost, constraints and reference profiles.An optimal energy management for the microgrid is challenging with respect to classical control theories. Needless to say, a DC microgrid is a complex system due to its heterogeneity, distributed nature (both spatial and in sampling time), nonlinearity of dynamics, multi-physic characteristics, the presence of constraints and uncertainties. Moreover, the power-preserving structure and the energy conservation of a microgrid are essential for ensuring a reliable operation.This challenges are tackled through the combined use of port-Hamiltonian formulations, differential flatness, and economic Model Predictive Control.The Port-Hamiltonian formalism allows to explicitly describe the power-preserving structure and the energy conservation of the microgrid and to connect different components of different physical natures through the same formalism. The strongly non-linear system is then translated into a flat representation. Taking into account differential flatness properties, reference profiles are generated such that the dissipated energy and various physical constraints are taken into account. Lastly, we minimize the purchasing/selling electricity cost within the microgrid using the economic Model Predictive Control with the Port-Hamiltonian formalism on graphs.The proposed control designs are validated through simulation results.

Lors du contrôle de robots, les variations fortes et soudaines dans les couples de commande doivent impérativement être évitées. En effet ces discontinuités peuvent entraîner, en plus des comportements imprévisibles du système, des dommages physiques, notamment au niveau des actionneurs. Pour la réalisation de tâches complexes, un robot à plusieurs degrés de liberté utilise généralement un système de commande multi-objectif avec lequel plusieurs tâches doivent être réalisées et plusieurs contraintes respectées. Le basculement entre ces différentes tâches ainsi que les contraintes causées par un environnemt dynamique et imprévisible sont les causes directes des variations fortes dans les couples de commande. Dans ce travail, les problèmes de transitions de priorités entre les différentes tâches ainsi que la variation des contraintes sont considérées avec pour objectif la des variations fortes dans les couples de commande. Deux contributions principales ont été réalisées.Premièrement, un nouveau contrôleur appelé "contrôle hiérarchique généralisé (GHC)" est implémenté sous forme d’optimisation quadratique pour gérer la priorité des transitions entre les tâches de poids différents. Le projecteur utilisé assure en plus de la continuité des transitions, la gestion de l’ajout et/ou de la suppression de tâches. Les couples de commande sont alors calculés en résolvant un problème d’optimisation prenant en compte en même temps la hiérarchie des tâches et les contraintes égalitaire et inégalitaires.Deuxièmement, nous avons développé une primitive de contrôle à base de Contrôle par Modèle Prédictif (CMP) afin de gérer l’existence des discontinuités des contraintes que doit respecter le robot, tel que le changement d’état des contacts ou l’évitement d'obstacles. Le contrôleur profite ainsi de la formulation prédictive en anticipant l'évolution des contraintes vis-à-vis des scénarios de commande et/ou de l'information des capteurs. Il permet de générer des nouvelles contraintes continues qui remplacent les anciennes contraintes discrètes dans le contrôleur réactif QP. Par conséquent, le taux de changement des couples articulaires est minimisé, comparé aux anciennes contraintes discrètes. Cette primitive de contrôle prédictive ne modifie pas directement les objectifs désirés des tâches mais les contraintes, ce qui permet de s’assurer que les changements de couple sont bien gérés dans les pires scénarios.L'efficacité de la stratégie de contrôle proposée est validée via des expériences en simulation avec le robot Kuka LWR 4+ et le robot humanoïde iCub. Les résultats montrent que l'approche développée peut réduire de manière significative la variation des couples articulaires pendant les changements de priorité des tâches ou sous contraintes discrètes. / Large and sudden changes in the torques of the actuators of a robot are highly undesirable and should be avoided during robot control as they may result in unpredictable behaviours. Multi-objective control system for complex robots usually have to handle multiple prioritized tasks while satisfying constraints. Changes in tasks and/or constraints are inevitable for robots when adapting to the unstructured and dynamic environment, and they may lead to large sudden changes in torques. Within this work, the problem of task priority transitions and changing constraints is primarily considered to reduce large sudden changes in torques. This is achieved through two main contributions as follows. Firstly, based on quadratic programming (QP), a new controller called Generalized Hierarchical Control (GHC) is developed to deal with task priority transitions among arbitrary prioritized task. This projector can be used to achieve continuous task priority transitions, as well as insert or remove tasks among a set of tasks to be performed in an elegant way. The control input (e.g. joint torques) is computed by solving one quadratic programming problem, where generalized projectors are adopted to maintain a task hierarchy while satisfying equality and inequality constraints. Secondly, a predictive control primitive based on Model Predictive Control (MPC) is developed to handle presence of discontinuities in the constraints that the robot must satisfy, such as the breaking of contacts with the environment or the avoidance of an obstacle. The controller takes the advantages of predictive formulations to anticipate the evolutions of the constraints by means of control scenarios and/or sensor information, and thus generate new continuous constraints to replace the original discontinuous constraints in the QP reactive controller. As a result, the rate of change in joint torques is minimized compared with the original discontinuous constraints. This predictive control primitive does not directly modify the desired task objectives, but the constraints to ensure that the worst case of changes of torques is well-managed. The effectiveness of the proposed control framework is validated by a set of experiments in simulation on the Kuka LWR robot and the iCub humanoid robot. The results show that the proposed approach significantly decrease the rate of change in joint torques when task priorities switch or discontinuous constraints occur.

Les centaines de millions de véhicules du parc automobile mondial nous rappellent à quel point notre société dépend du moteur à combustion interne. Malgré des progrès significatifs en termes d’émissions polluantes et de consommation, les moteurs à essence et diesel demeurent l’une des principales sources de pollution de l’air des centres urbains modernes. Ce constat motive les autorités à renforcer les normes anti-pollution, qui tendent à complexifier la définition technique des moteurs. En particulier, un nombre croissant d’actionneurs fait aujourd’hui, du contrôle de la chaîne d’air, un challenge majeur. Dans un marché de plus en plus mondialisé et où le temps de développement de moteurs se doit d’être de plus en plus court, ces travaux entendent proposer une solution aux problèmes liés à cette augmentation de la complexité. La proposition repose sur une approche en trois étapes et combine : modélisation physique du moteur, contrôle prédictif non linéaire et programmation multiparamétrique. Le cas du contrôle de la chaîne d’air d’un moteur à essence suralimenté sert de fil conducteur au document. Dans son ensemble, les développements présentés ici fournissent une approche quasi-systématique pour la synthèse du contrôle de la chaîne des moteurs à essence suralimentés. Intuitivement, le raisonnement doit pouvoir être étendu à d’autres boucles de contrôle et au cas des moteurs diesel. / The hundreds of millions of passenger cars and other vehicles on our roads emphasize our society’s reliance on internal combustion engines. Despite striking progress in terms of pollutant emissions and fuel consumption, gasoline and diesel engines remain one of the most important sources of air pollution in modern urban areas. This leads the authorities to lay down increasingly drastic pollutant emission standards, which entail ever more complex engine technical definitions. In particular, due to an increasing number of actuators in the past few years, the air path of internal combustion engines represents one of the biggest challenges of engine control design. The present thesis addresses this issue of increasing engine complexity with respect to the continuous reduction in development time, dictated by a more and more competitive globalized market. The proposal consists in a three-step approach that combines physics-based engine modeling, nonlinear model predictive control and multi-parametric nonlinear programming. The latter leads to an explicit piecewise affine feedback control law, compatible with a real-time implementation. The proposed approach is applied to the particular case of the control of the air path of a turbocharged gasoline engine. Overall, the developments presented in this thesis provide a quasi-systematic approach for the synthesis of the control of the air path of turbocharged gasoline engines. Intuitively, this approach can be extended to other control loops in both gasoline and diesel engines.

Doctor of Philosophy / Electrical and Computer Engineering / Balasubramaniam Natarajan / A cyber-physical system (CPS) typically consists of networked computational elements that control physical processes. As an integral part of CPS, the widespread deployment of communicable sensors makes the task of monitoring and control quite challenging especially from the viewpoint of scalability and complexity. This research investigates two unique aspects of overcoming such barriers, making a CPS more robust against data explosion and network vulnerabilities. First, the correlated characteristics of high-resolution sensor data are exploited to significantly reduce the fused data volume. Specifically, spatial, temporal and spatiotemporal compressed sensing approaches are applied to sample the measurements in compressed form. Such aggregation can directly be used in centralized static state estimation even for a nonlinear system. This approach results in a remarkable reduction in communication overhead as well as memory/storage requirement. Secondly, an agent based architecture is proposed, where the communicable sensors (identified as agents) also perform local information processing. Based on the local and underdetermined observation space, each agent can monitor only a specific subset of global CPS states, necessitating neighborhood information exchange. In this framework, we propose an agent based static state estimation encompassing local consensus and least square solution. Necessary bounds for the consensus weights are obtained through the maximum eigenvalue based convergence analysis and are verified for a radial power distribution network. The agent based formulation is also applied for a linear dynamical system and the consensus approach is found to exhibit better and more robust performance compared to a diffusion filter. The agent based Kalman consensus filter (AKCF) is further investigated, when the agents can choose between measurements and/or consensus, allowing the economic allocation of sensing and communication tasks as well as the temporary omission of faulty agents. The filter stability is guaranteed by deriving necessary consensus bounds through Lyapunov stability analysis. The states dynamically estimated from AKCF can be used for state-feedback control in a model predictive fashion. The effect of lossy communication is investigated and critical bounds on the link failure rate and the degree of consensus that ensure stability of the agent based control are derived and verified via simulations.

abstract: Control engineering offers a systematic and efficient approach to optimizing the effectiveness of individually tailored treatment and prevention policies, also known as adaptive or ``just-in-time'' behavioral interventions. These types of interventions represent promising strategies for addressing many significant public health concerns. This dissertation explores the development of decision algorithms for adaptive sequential behavioral interventions using dynamical systems modeling, control engineering principles and formal optimization methods. A novel gestational weight gain (GWG) intervention involving multiple intervention components and featuring a pre-defined, clinically relevant set of sequence rules serves as an excellent example of a sequential behavioral intervention; it is examined in detail in this research.   A comprehensive dynamical systems model for the GWG behavioral interventions is developed, which demonstrates how to integrate a mechanistic energy balance model with dynamical formulations of behavioral models, such as the Theory of Planned Behavior and self-regulation. Self-regulation is further improved with different advanced controller formulations. These model-based controller approaches enable the user to have significant flexibility in describing a participant's self-regulatory behavior through the tuning of controller adjustable parameters. The dynamic simulation model demonstrates proof of concept for how self-regulation and adaptive interventions influence GWG, how intra-individual and inter-individual variability play a critical role in determining intervention outcomes, and the evaluation of decision rules.   Furthermore, a novel intervention decision paradigm using Hybrid Model Predictive Control framework is developed to generate sequential decision policies in the closed-loop. Clinical considerations are systematically taken into account through a user-specified dosage sequence table corresponding to the sequence rules, constraints enforcing the adjustment of one input at a time, and a switching time strategy accounting for the difference in frequency between intervention decision points and sampling intervals. Simulation studies illustrate the potential usefulness of the intervention framework. The final part of the dissertation presents a model scheduling strategy relying on gain-scheduling to address nonlinearities in the model, and a cascade filter design for dual-rate control system is introduced to address scenarios with variable sampling rates. These extensions are important for addressing real-life scenarios in the GWG intervention. / Dissertation/Thesis / Doctoral Dissertation Chemical Engineering 2014

The global greenhouse gas CO2 emission from the transportation sector is very significant. To reduce this gas emission, EU has set an average target of not more than 95 CO2/km for new passenger cars by the year 2020. A great reduction is still required to achieve the CO2 emission target in 2020, and many different approaches are being considered. This thesis focuses on the thermal management of the engine as an area that promise significant improvement of fuel efficiency with relatively small changes. The review of the literature shows that thermal management can improve engine efficiency through the friction reduction, improved air-fuel mixing, reduced heat loss, increased engine volumetric efficiency, suppressed knock, reduce radiator fan speed and reduction of other toxic emissions such as CO, HC and NOx. Like heat loss and friction, most emissions can be reduced in high temperature condition, but this may lead to poor volumetric efficiency and make the engine more prone to knock. The temperature trade-off study is conducted in simulation using a GT-SUITE engine model coupled with the FE in-cylinder wall structure and cooling system. The result is a map of the best operating temperature over engine speed and load. To quantify the benefit of this map, eight driving styles from the legislative and research test cycles are being compared using an immediate application of the optimal temperature, and significant improvements are found for urban style driving, while operation at higher load (motorway style driving) shows only small efficiency gains. The fuel consumption saving predicted in the urban style of driving is more than 4%. This assess the chance of following the temperature set point over a cycle, the temperature reference is analysed for all eight types of drive cycles using autocorrelation, lag plot and power spectral density. The analysis consistently shows that the highest volatility is recorded in the Artemis Urban Drive Cycle: the autocorrelation disappears after only 5.4 seconds, while the power spectral density shows a drop off around 0.09Hz. This means fast control action is required to implement the optimal temperature before it changes again. Model Predictive Control (MPC) is an optimal controller with a receding horizon, and it is well known for its ability to handle multivariable control problems for linear systems with input and state limits. The MPC controller can anticipate future events and can take control actions accordingly, especially if disturbances are known in advance. The main difficulty when applying MPC to thermal management is the non-linearity caused by changes in flow rate. Manipulating both the water pump and valve improves the control authority, but it also amplifies the nonlinearity of the system. Common linearization approaches like Jacobian Linearization around one or several operating points are tested, by found to be only moderately successful. Instead, a novel approach is pursued using feedback linearization of the plant model. This uses an algebraic transformation of the plant inputs to turn the nonlinear systems dynamics into a fully or predominantly linear system. The MPC controller can work with the linear model, while the actual control inputs are found using an inverse transformation. The Feedback Linearization MPC of the cooling system model is implemented and testing using MathWork Simulink®. The process includes the model transformation approach, model fitting, the transformation of the constraints and the tuning of the MPC controller. The simulation shows good temperature tracking performance, and this demonstrates that a MPC controller with feedback linearization is a suitable approach to thermal management. The controller strategy is then validated in a test rig replicating an actual engine cooling system. The new MPC controller is again evaluated over the eight driving cycles. The average water pump speed is reduced by 9.1% compared to the conventional cooling system, while maintaining good temperature tracking. The controller performance further improves with future disturbance anticipation by 20.5% for the temperature tracking (calculated by RMSE), 6.8% reduction of the average water pump speed, 47.3% reduction of the average valve movement and 34.0% reduction of the average radiator fan speed.

abstract: Increasing interest in individualized treatment strategies for prevention and treatment of health disorders has created a new application domain for dynamic modeling and control. Standard population-level clinical trials, while useful, are not the most suitable vehicle for understanding the dynamics of dosage changes to patient response. A secondary analysis of intensive longitudinal data from a naltrexone intervention for fibromyalgia examined in this dissertation shows the promise of system identification and control. This includes datacentric identification methods such as Model-on-Demand, which are attractive techniques for estimating nonlinear dynamical systems from noisy data. These methods rely on generating a local function approximation using a database of regressors at the current operating point, with this process repeated at every new operating condition. This dissertation examines generating input signals for data-centric system identification by developing a novel framework of geometric distribution of regressors and time-indexed output points, in the finite dimensional space, to generate sufficient support for the estimator. The input signals are generated while imposing “patient-friendly” constraints on the design as a means to operationalize single-subject clinical trials. These optimization-based problem formulations are examined for linear time-invariant systems and block-structured Hammerstein systems, and the results are contrasted with alternative designs based on Weyl's criterion. Numerical solution to the resulting nonconvex optimization problems is proposed through semidefinite programming approaches for polynomial optimization and nonlinear programming methods. It is shown that useful bounds on the objective function can be calculated through relaxation procedures, and that the data-centric formulations are amenable to sparse polynomial optimization. In addition, input design formulations are formulated for achieving a desired output and specified input spectrum. Numerical examples illustrate the benefits of the input signal design formulations including an example of a hypothetical clinical trial using the drug gabapentin. In the final part of the dissertation, the mixed logical dynamical framework for hybrid model predictive control is extended to incorporate a switching time strategy, where decisions are made at some integer multiple of the sample time, and manipulation of only one input at a given sample time among multiple inputs. These are considerations important for clinical use of the algorithm. / Dissertation/Thesis / Ph.D. Electrical Engineering 2014

The main topic of this thesis is model predictive control (MPC) of an unstable fighter aircraft. When flying it is important to be able to reach, but not exceed the aircraft limitations and to consider the physical boundaries on the control signals. MPC is a method for controlling a system while considering constraints on states and control signals by formulating it as an optimization problem. The drawback with MPC is the computational time needed and because of that, it is primarily developed for systems with a slowly varying dynamics. Two different methods are chosen to speed up the process by making simplifications, approximations and exploiting the structure of the problem. The first method is an explicit method, performing most of the calculations offline. By solving the optimization problem for a number of data sets and thereafter training a neural network, it can be treated as a simpler function solved online. The second method is called fast MPC, in this case the entire optimization is done online. It uses Cholesky decomposition, backward-forward substitution and warm start to decrease the complexity and calculation time of the program. Both methods perform reference tracking by solving an underdetermined system by minimizing the weighted norm of the control signals. Integral control is also implemented by using a Kalman filter to observe constant disturbances. An implementation was made in MATLAB for a discrete time linear model and in ARES, a simulation tool used at Saab Aeronautics, with a more accurate nonlinear model. The result is a neural network function computed in tenth of a millisecond, a time independent of the size of the prediction horizon. The size of the fast MPC problem is however directly affected by the horizon and the computational time will never be as small, but it can be reduced to a couple of milliseconds at the cost of optimality.

A identificação de modelos é usualmente a tarefa mais significativa e demorada no trabalho de implementação e manutenção de sistemas de controle que usam Controle Preditivo baseado em Modelos (MPC) tendo em vista a complexidade da tarefa e a importância que o modelo possui para um bom desempenho do controlador. Após a implementação, o controlador tende a permanecer com o modelo original mesmo que mudanças de processo tenham ocorrido levando a uma degradação das ações do controlador. Este trabalho apresenta uma aplicação industrial de re-identificação em malha fechada. A metodologia de excitação da planta utilizada foi apresentada em Sotomayor et al. (2009). Tal técnica permite obter o comportamento das variáveis de processo sem desligar o MPC e sem modificar sua estrutura, aumentando assim, o automatismo e a segurança do procedimento de re-identificação. O sistema re-identificado foi uma coluna debutanizadora de uma refinaria brasileira sendo que os modelos fazem parte do controle preditivo multivariável dessa coluna de destilação. A metodologia foi aplicada com sucesso podendo-se obter os seis novos modelos para atualizar o controlador em questão, o que resultou em uma melhoria de seu desempenho. / Model identification is usually the most significant and time-consuming task of implementing and maintaining control systems based on models (MPC) concerning the complexity of the task and the importance of the model for a good performance of the controller. After being implemented the MPC tends to remain with the original model even after process changes have occurred, leading to a degradation of the controller actions. The present work shows an industrial application of closed-loop re-identification. The plant excitation methodology used here was presented in Sotomayor et al. (2009). Such technique allows for obtaining the behavior of the process variables with the MPC still working and without modifying the MPC structure, increasing automation and safety of the re-identification procedure. The system re-identified was a debutanizer column of a Brazilian refinery being the models part of the multivariable predictive control of this distillation column. The methodology was applied with reasonable success managing to obtain 6 new models to update this MPC, and resulting in improved control performance.