Dissertations / Theses on the topic 'Networked Model Predictive Control'

We present a motion control framework which allows a group of robots to work together to decide upon their motions by minimizing a collective cost without any central computing component or any one agent performing a large portion of the computation. When developing distributed control algorithms, care must be taken to respect the limited computational capacity of each agent as well as respect the information and communication constraints of the network. To address these issues, we develop a distributed, behavior-based model predictive control (MPC) framework which alleviates the computational difficulties present in many distributed MPC frameworks, while respecting the communication and information constraints of the network. In developing the multi-agent control framework, we make three contributions. First, we develop a distributed optimization technique which respects the dynamic communication restraints of the network, converges to a collective minimum of the cost, and has transients suitable for robot motion control. Second, we develop a behavior-based MPC framework to control the motion of a single-agent and apply the framework to robot navigation. The third contribution is to combine the concepts of distributed optimization and behavior-based MPC to develop the mentioned multi-agent behavior-based MPC algorithm suitable for multi-robot motion control.

More and more renewable energy sources are being integrated into microgrids—and while this causes many control challenges for microgrids, it can also yield numerous economic and environmental benefits. Therefore, it is necessary to develop proper control schemes for microgrids to address the different control issues in their hierarchical structure while adapting to the different time scales of the three control levels. Conversely, because model predictive control (MPC) has significant advantages—the inclusion of forecasts, the simplicity of the algorithm, and the flexibility to handle hard constraints—it has attracted significant attention in industrial control systems. Motivated by these factors, this research focuses on implementing MPC techniques in microgrids, which are solely supplied by photovoltaic (PV) generators, to address different control problems. For primary control of the microgrid hierarchy, which is mainly responsible for the inner control of the local distributed generation units, MPC can be applied to control of the power converters that serve as interfaces between the sources and the loads. Therefore, in this control level, a novel output-feedback MPC technique based on ellipsoidal set-membership state estimation is designed for a direct current to direct current (DC-DC) converter, considering the unknown-but-bounded external disturbances. A long-horizon finite-states (FS) MPC strategy is designed for the direct current to alternating current (DC-AC) inverter to reduce the sampling and switching frequency through a multi-step implementation approach and a control sequence rearrangement method. For secondary control, which is in charge of the compensation for the frequency and voltage deviations and is usually communication-based, the distributed MPC strategy can be used to realize the desired cooperative control among the geographically dispersed units. Thus, a novel distributed model predictive controller is developed to enhance system performance. It takes into account the fact that the distributed controllers’ communication network might be subject to switching topology due to the disconnection and reconnection of controllers, random failures, and recoveries of the links between controllers. A Markov chain with a time-varying probability transition matrix is used to describe the stochastic topology evolution of the control network. Tertiary control is used to coordinate the power flow between the microgrid and the utility grid and offers economic operations for microgrids. Since the integration of renewable energy sources causes low inertia and power fluctuation in microgrids, battery energy storage is essential to addressing these issues. To coordinate the charging/discharging schedule of the battery storage units, a networked MPC strategy can be adopted to realize the communication between different microgrid components and make use of the forecasts for PV power generation and load demand. The multi-microgrid system is considered subject to partial fault because of non-functional generators, batteries, or even transmission lines in this research. Hence, both the connection status of the electrical grid and the communication network are incorporated into the system modeling. In addition, the set-membership estimation is adopted to deal with the possible state unavailability caused by non-functional batteries or communication failures. In the theoretical section of this thesis, different sufficient conditions are established to ensure the stability of the investigated systems, and the optimal control inputs are obtained by solving the corresponding optimization problems. For easy implementation with MATLAB solvers, all the constraints and conditions of the optimization problems are transformed into linear matrix inequalities. Different recursive MPC algorithms are designed to control the target systems, and some extended algorithms are also developed to assist with the computation to determine the optimal solutions. In the demonstration section of this thesis, the designed controllers are all implemented in the numerical simulations or Simulink tests to verify their effectiveness, and an experimental test based on Raspberry Pi is conducted to demonstrate the wireless communication employing the designed networked model predictive controller.
Thesis (PhD Doctorate)
Doctor of Philosophy (PhD)
School of Eng & Built Env
Science, Environment, Engineering and Technology
Full Text

Wireless networks in industrial process control enable new system architectures and designs. However, wireless control systems can be severely affected by the imperfections of the communication links. This thesis proposes new methods to handle such imperfections by adding additional components in the control loop, or by adapting sampling intervals and control actions. First, the predictive outage compensator is proposed. It is a filter which is implemented at the receiver side of networked control systems. There it generates predicted samples when data are lost, based on past data. The implementation complexity of the predictive outage compensator is analyzed. Simulation and experimental results show that it can considerably improve the closed-loop control performance under communication losses. The thesis continues with presenting an algorithm for controlling multiple processes on a shared communication network, using adaptive sampling intervals. The methodology is based on model predictive control, where the controller jointly decides the optimal control signal to be applied as well as the optimal time to wait before taking the next sample. The approach guarantees conflict-free network transmissions for all controlled processes. Simulation results show that the presented control law reduces the required amount of communication, while maintaining control performance. The third contribution of the thesis is an event-triggered model predictive controller for use over a wireless link. The controller uses open-loop optimal control, re-computed and communicated only when the system behavior deviates enough from a prediction. Simulations underline the methods ability to significantly reduce computation and communication effort, while guaranteeing a desired level of system performance. The thesis is concluded by an experimental validation of wireless control for a physical lab process. A hybrid model predictive controller is used, connected to the physical system through a wireless medium. The results reflect the advantages and challenges in wireless control.

QC 20140217

2009/2010
One of the more recent and promising approaches to control is the Receding Horizon one. Due to its intrinsic characteristics, this methodology, also know as Model Predictive Control, allows to easily face disturbances and model uncertainties: indeed at each sampling instant the control action is recalculated on the basis of the reached state (closed loop). More in detail, the procedure consists in the minimization of an adequate cost function with respect to a control input sequence; then the first element of the optimal sequence is applied. The whole procedure is then continuously reiterated. In this thesis, we will focus in particular on robust control of constrained systems. This is motivated by the fact that, in practice, every real system is subjected to uncertainties, disturbances and constraints, in particular on state and input (for instance, plants can work without being damaged only in a limited set of configurations and, on the other side, control actions must be compatible with actuators' physical limits). With regard to the first aspect, maintaining the closed loop stability even in presence of disturbances or model mismatches can result in an essential strategy: moreover it can be exploited in order to design an approximate stabilizing controller, as it will be shown. The control input values are obtained recurring to a Nearest Neighbour technique or, in more favourable cases, to a Neural Network based approach to the exact RH law, which can be then calculated off line: this implies a strong improvement related to the applicability of MPC policy in particular in terms of on line computational burden. The proposed scheme is capable to guarantee stability even for systems that are not stabilizable by means of a continuous feedback control law. Another interesting framework in which the study of the influence of uncertainties on stability can lead to significant contributions is the networked MPC one. In this case, due to the absence of physical interconnections between the controller and the systems to be controlled, stability can be obtained only taking into account of the presence of disturbances, delays and data losses: indeed this kind of uncertainties are anything but infrequent in a communication network. The analysis carried out in this thesis regards interconnected systems and leads to two distinct procedures, respectively stabilizing the linear systems with TCP protocol and nonlinear systems with non-acknowledged protocol. The core of both the schemes resides in the online solution of an adequate reduced horizon optimal control problem.
Una delle strategie di controllo emerse più recentemente, più promettenti e di conseguenza più studiate negli ultimi anni è quella basata sull'approccio Receding Horizon. Grazie alle caratteristiche che contraddistinguono questa tecnica, cui si fa spesso riferimento anche col nome di Model Predictive Control, risulta piuttosto agevole trattare eventuali disturbi e incertezze di modellazione; tale metodo prevede infatti il calcolo di un nuovo ingresso di controllo per ciascun istante di campionamento, in seguito alla minimizzazione ad ogni passo di un'opportuna funzione di costo rispetto ad una sequenza di possibili futuri ingressi, inizializzata sulla base del valore dello stato del sistema all'istante considerato. Il controllo è dato dal primo elemento di tale sequenza ottima; tutto questo viene continuamente ripetuto, il che comporta un aggiornamento costante del segnale di controllo. Gli inconvenienti di questa tecnica risiedono nelle elevate risorse computazionali e nei tempi di calcolo richiesti, così da ridurne drasticamente l'applicabilità specie nel caso di sistemi con elevata dinamica. In questa tesi ci si concentrerà sulle caratteristiche di robustezza del controllore: l'importanza di quest'analisi risiede nel fatto che ogni sistema reale è soggetto a incertezze e disturbi di varia origine cui bisogna far fronte durante le normali condizioni di funzionamento. Inoltre, la capacità di gestire errori di modellazione, come si vedrà, può essere sfruttata per ottenere un notevole incremento delle prestazioni nella stima del valore da fornire in ingresso all'impianto: si tratta di ripartire l'errore complessivo in modo da garantirsi dei margini che consentano di lavorare con un'approssimazione della legge di controllo, come specificato più avanti. In tutto il lavoro si considereranno sistemi vincolati: l'interesse per questa caratteristica dipende dal fatto che nella pratica vanno sempre tenuti in considerazione eventuali vincoli su stato e ingressi: basti pensare al fatto che ogni impianto è progettato per lavorare solo all'interno un determinato insieme di configurazioni, determinato ad esempio da vincoli fisici su attuatori, sensori e così via: non riporre sufficiente attenzione in tali restrizioni può risultare nel danneggiamento del sistema di controllo o dell'impianto stesso. Le caratteristiche di stabilità di un sistema controllato mediante MPC dipendono in modo determinante dalla scelta dei parametri e degli attributi della funzione di costo da minimizzare; nel seguito, con riferimento al caso dei sistemi non lineari, saranno forniti suggerimenti e strumenti utili in tal senso, al fine di ottenere la stabilità anche in presenza di disturbi (che si assumeranno opportunamente limitati). Successivamente tale robustezza verrà sfruttata per la progettazione di controllori stabilizzanti approssimati: si dimostrerà infatti che, una volta progettato adeguatamente il sistema di controllo “esatto” basato su approccio RH e conseguentemente calcolati off-line i valori ottimi degli ingressi su una griglia opportunamente costruita sul dominio dello stato, il ricorso a una conveniente approssimazione di tali valori non compromette le proprietà di stabilità del sistema complessivo, che continua per di più a mantenere una certa robustezza. Da notare che ciò vale anche per sistemi non stabilizzabili mediante legge di controllo feedback continua: la funzione approssimante può essere ottenuta in questo caso con tecniche di tipo Nearest Neighbour; qualora invece la legge di controllo sia sufficientemente regolare si potrà far ricorso ad approssimatori smooth, quali ad esempio le reti neurali. Tutto ciò comporta un notevole miglioramento delle prestazioni del controllore RH sia dal punto di vista del tempo di calcolo richiesto che (nel secondo caso) della memoria necessaria ad immagazzinare i parametri del controllore, risultando nell'applicabilità dell'approccio basato su MPC anche al caso di sistemi con elevata dinamica. Un altro ambito in cui lo studio dell'influenza delle incertezze e dei disturbi sulla stabilità richiede una notevole attenzione è quello dei sistemi networked; anche in questo caso il ricorso all'MPC può portare a ottimi risultati di stabilità robusta, a patto di individuare un' opportuna struttura per il sistema complessivo ed effettuare scelte adeguate per il problema di ottimizzazione. In particolare, si considererà il caso di trasmissione di dati tra un controllore centralizzato e le varie parti dell'impianto in assenza di collegamento fisico diretto. Lo studio della stabilità dovrà allora tenere in considerazione la presenza di perdite di pacchetti o ritardi di trasmissione, condizioni tutt'altro che infrequenti per le reti. Saranno quindi proposte due distinte procedure, che si dimostreranno essere in grado di garantire robustezza a sistemi rispettivamente lineari comunicanti con protocolli di tipo TCP e non lineari in presenza di protocolli UDP. Questo secondo caso è senz'altro il più complesso ma allo stesso tempo il più concreto tra i due. Il nucleo del controllo è ancora basato su una tecnica MPC, ma stavolta il controllore è chiamato a risolvere il problema di ottimizzazione su un orizzonte “ridotto”, che consente la gestione dei ritardi e di eventuali perdite di pacchetto su determinati canali. La lunghezza dell'orizzonte dipenderà dalla presenza o meno dei segnali di ricezione del pacchetto (acknowledgement).
XXIII Ciclo
1977

Model Predictive Control (MPC) is an optimization-based paradigm forfeedback control. The MPC relies on a dynamical model to make predictionsfor the future values of the controlled variables of the system. It then solvesa constrained optimization problem to calculate the optimal control actionthat minimizes the difference between the predicted values and the desiredor set values. One of the main limitations of the traditional MPC lies in thehigh computational cost resulting from solving the associated optimizationproblem online. Various offline strategies have been proposed to overcomethis, ranging from the explicit MPC (eMPC) to the recent learning-basedneural network approaches. This thesis investigates a framework for thetraining and evaluation of a neural network for learning to implement theMPC. As a part of the framework, a new approach for efficient generationof training data is proposed. Four different neural network structures arestudied; one of them is a black box network while the other three employMPC specific information. The networks are evaluated in terms of twodifferent performance metrics through experiments conducted on realistictwo-dimensional and four-dimensional systems. The experiments revealthat while using MPC specific structure in the neural networks resultsin performance gains when the training data is limited, all the networkstructures perform similarly as extensive training data is used. They furthershow that a recurrent neural network structure trained on both the state andcontrol trajectories of a family of MPCs is able to generalize to previouslyunseen MPC problems. The proposed methods in this thesis act as a firststep towards developing a coherent framework for characterization of learningapproaches in terms of both model validation and efficient training datageneration.
Modell-prediktiv reglering (MPC) är en strategi inom återkopplad regleringmed rötter i optimeringsteori. MPC:n använder sig av en dynamiskmodell för att prediktera de framtida värdena på systemets styrvariabler.Den löser sedan ett optimeringsproblem för att beräkna en optimalstyrsignal som minimerar skillnaden mellan referensvärdena och depredikterade värdena. Att lösa det associerade optimeringsproblemetonline kan medföra höga beräkningskostnader, något som utgör en av dehuvudsakliga begränsningarna med traditionell MPC. Olika offline-strategierhar föreslagits för att kringgå detta, däribland explicit modell-prediktivreglering (eMPC) samt senare inlärningsmetoder baserade på neuronnät.Den här masteruppsatsen undersöker ett ramverk för träning och utvärderingav olika neuronnätsstrukturer för MPC-inlärning. En ny metod för effektivgenerering av träningsdata presenteras som en del av detta ramverk.Fyra olika nätstrukturer studeras; ett black box-nät samt tre nät sominkluderar MPC-specifik information. Näten evalueras i termer av två olikaprestandamått genom experiment på realistiska två- och fyrdimensionellasystem. Experimenten visar att en MPC-specifik nätstruktur resulterar iökad prestanda när mängden träningsdata är begränsad, men att de fyranäten presterar likvärdigt när mycket träningsdata finns att tillgå. De visarvidare att ett återkopplat neuronnät som tränas på både tillstånds- ochstyrsignalstrajektorier från en familj av MPC:er har förmågan att generaliseravid påträffandet av nya MPC-problem. De föreslagna metoderna i den häruppsatsen utgör ett första steg mot utvecklandet av ett enhetligt ramverk förkaraktärisering av inlärningsmetoder i termer av både modellvalidering ocheffektiv datagenerering.

2007/2008
Nonlinear Receding Horizon (RH) control, also known as moving horizon control or nonlinear Model Predictive Control (MPC), refers to a class of algorithms that make explicit use of a nonlinear process model to optimize the plant behavior, by computing a sequence of future ma- nipulated variable adjustments. Usually the optimal control sequence is obtained by minimizing a multi-stage cost functional on the basis of open-loop predictions. The presence of uncertainty in the model used for the optimization raises the question of robustness, i.e., the maintenance of certain properties such as stability and performance in the presence of uncertainty. The need for guaranteeing the closed-loop stability in presence of uncertainties motivates the conception of robust nonlinear MPC, in which the perturbations are explicitly taken in account in the design of the controller. When the nature of the uncertainty is know, and it is assumed to be bounded in some compact set, the robust RH control can be determined, in a natural way, by solving a min–max optimal control problem, that is, the performance objective is optimized for the worst-case scenario. However, the use of min-max techniques is limited by the high computational burden required to solve the optimization problem. In the case of constrained system, a possibility to ensure the robust constraint satisfaction and the closed-loop stability without resorting to min-max optimization consists in imposing restricted (tightened) constraints on the the predicted trajectories during the optimization. In this framework, an MPC scheme with constraint tightening for discrete-time nonlinear systems affected by state-dependent and norm bounded uncertainties is proposed and discussed. A novel method to tighten the constraints relying on the nominal state prediction is described, leading to less conservative set contractions than in the existing approaches. Moreover, by imposing a stabilizing state constraint at the end of the control horizon (in place of the usual terminal one placed at the end of the prediction horizon), less stringent assumptions can be posed on the terminal region, while improving the robust stability properties of the MPC closed-loop system. The robust nonlinear MPC formulation with tightened constraints is then used to design off- line approximate feedback laws able to guarantee the practical stability of the closed-loop system. By using off-line approximations, the computational burden due to the on-line optimization is removed, thus allowing for the application of the MPC to systems with fast dynamics. In this framework, we will also address the problem of approximating possibly discontinuous feedback functions, thus overcoming the limitation of existent approximation scheme which assume the continuity of the RH control law (whereas this condition is not always verified in practice, due to both nonlinearities and constraints). Finally, the problem of stabilizing constrained systems with networked unreliable (and de- layed) feedback and command channels is also considered. In order to satisfy the control ob- jectives for this class of systems, also referenced to as Networked Control Systems (NCS’s), a control scheme based on the combined use of constraint tightening MPC with a delay compen- sation strategy will be proposed and analyzed. The stability properties of all the aforementioned MPC schemes are characterized by using the regional Input-to-State Stability (ISS) tool. The ISS approach allows to analyze the depen- dence of state trajectories of nonlinear systems on the magnitude of inputs, which can represent control variables or disturbances. Typically, in MPC the ISS property is characterized in terms of Lyapunov functions, both for historical and practical reasons, since the optimal finite horizon cost of the optimization problem can be easily used for this task. Note that, in order to study the ISS property of MPC closed-loop systems, global results are in general not useful because, due to the presence of state and input constraints, it is impossible to establish global bounds for the multi-stage cost used as Lyapunov function. On the other hand local results do not allow to analyze the properties of the predictive control law in terms of its region of attraction. There- fore, regional ISS results have to employed for MPC controlled systems. Moreover, in the case of NCS, the resulting control strategy yields to a time-varying closed-loop system, whose stability properties can be analyzed using a novel regional ISS characterization in terms of time-varying Lyapunov functions.
XXI Ciclo
1980

Although nonlinear model predictive control (NMPC) might be the best choice for a nonlinear plant, it is still not widely used. This is mainly due to the computational burden associated with solving online a set of nonlinear differential equations and a nonlinear dynamic optimization problem in real time. This thesis is concerned with strategies aimed at reducing the computational burden involved in different stages of the NMPC such as optimization problem, state estimation, and nonlinear model identification. A major part of the computational burden comes from function and derivative evaluations required in different parts of the NMPC algorithm. In this work, the problem is tackled using a recently introduced efficient tool, the automatic differentiation (AD). Using the AD tool, a function is evaluated together with all its partial derivative from the code defining the function with machine accuracy. A new NMPC algorithm based on nonlinear least square optimization is proposed. In a first–order method, the sensitivity equations are integrated using a linear formula while the AD tool is applied to get their values accurately. For higher order approximations, more terms of the Taylor expansion are used in the integration for which the AD is effectively used. As a result, the gradient of the cost function against control moves is accurately obtained so that the online nonlinear optimization can be efficiently solved. In many real control cases, the states are not measured and have to be estimated for each instance when a solution of the model equations is needed. A nonlinear extended version of the Kalman filter (EKF) is added to the NMPC algorithm for this purpose. The AD tool is used to calculate the required derivatives in the local linearization step of the filter automatically and accurately. Offset is another problem faced in NMPC. A new nonlinear integration is devised for this case to eliminate the offset from the output response. In this method, an integrated disturbance model is added to the process model input or output to correct the plant/model mismatch. The time response of the controller is also improved as a by–product. The proposed NMPC algorithm has been applied to an evaporation process and a two continuous stirred tank reactor (two–CSTR) process with satisfactory results to cope with large setpoint changes, unmeasured severe disturbances, and process/model mismatches. When the process equations are not known (black–box) or when these are too complicated to be used in the controller, modelling is needed to create an internal model for the controller. In this thesis, a continuous time recurrent neural network (CTRNN) in a state–space form is developed to be used in NMPC context. An efficient training algorithm for the proposed network is developed using AD tool. By automatically generating Taylor coefficients, the algorithm not only solves the differentiation equations of the network but also produces the sensitivity for the training problem. The same approach is also used to solve online the optimization problem of the NMPC. The proposed CTRNN and the predictive controller were tested on an evaporator and two–CSTR case studies. A comparison with other approaches shows that the new algorithm can considerably reduce network training time and improve solution accuracy. For a third case study, the ALSTOM gasifier, a NMPC via linearization algorithm is implemented to control the system. In this work a nonlinear state–space class Wiener model is used to identify the black–box model of the gasifier. A linear model of the plant at zero–load is adopted as a base model for prediction. Then, a feedforward neural network is created as the static gain for a particular output channel, fuel gas pressure, to compensate its strong nonlinear behavior observed in open–loop simulations. By linearizing the neural network at each sampling time, the static nonlinear gain provides certain adaptation to the linear base model. The AD tool is used here to linearize the neural network efficiently. Noticeable performance improvement is observed when compared with pure linear MPC. The controller was able to pass all tests specified in the benchmark problem at all load conditions.

Today's Cyber-Physical Systems (CPS) are typically distributed over several computing nodes communicating by way of shared buses such as Controller Area Network (CAN). Their control performance gets degraded due to variable delays (jitters) incurred by messages on the shared CAN bus due to contention and network overhead. This work presents a novel online delay prediction approach that predicts the message delay at runtime based on real-time traffic information on CAN. It leverages the proposed method to improve control quality, by compensating for the message delay using the Model Predictive Control (MPC) algorithm in designing the controller. By simulating an automotive Cruise Control system and a DC Motor plant in a CAN environment, it goes on to demonstrate that the delay prediction is accurate, and that the MPC design which takes the message delay into consideration, performs considerably better. It also implements the proposed method on an 8-bit 16MHz ATmega328P microcontroller and measures the execution time overhead. The results clearly indicate that the method is computationally feasible for online usage.
Master of Science

The aim of the thesis is to develop a model-based control strategy, namely, the Model Predictive Control (MPC) method, for robot position control using artificial neural networks. MPC is primarily developed for process control. Therefore its application in robot control has been less reported. In addition, conventional MPC uses linear model of the system for prediction which leads to inaccuracy for highly non-linear systems, such as robot. In this thesis a simulation model of a modified PUMA robot is constructed. This model is built using both MATLAB/SIMULINK and FORTRAN languages. In this model, the full robot dynamics is used together with the realistic factors, such as the actuator effects and the gear backlash, to represent the real system accurately. All simulations throughout this thesis are carried out on this model. A model predictive control strategy for robot trajectory tracking is also introduced in this thesis. The feasibility of the proposed MPC control method is studied based on a perfect prediction model, a model with uncertainties, and when the frequency band of the MPC controller is limited. Furthermore, a new method of using neural networks for robot dynamics modelling is introduced. This method is developed on the basis of a numerical differential technique that eliminates the explicit requirement of robot joint accelerations. Therefore, this method can be easily implemented on physical systems. As the measurements of the robot joint positions, velocities, and torques collected from operating the robot can be used to train the neural network, a more accurate dynamic model can be obtained. Finally, the MPC control method and the neural network model are combined together to form a neural network based MPC controller. The validity of this method is verified by using simulation on the simulated robot system
Master of Engineering (Hons)

The aim of the thesis is to develop a model-based control strategy, namely, the Model Predictive Control (MPC) method, for robot position control using artificial neural networks. MPC is primarily developed for process control. Therefore its application in robot control has been less reported. In addition, conventional MPC uses linear model of the system for prediction which leads to inaccuracy for highly non-linear systems, such as robot. In this thesis a simulation model of a modified PUMA robot is constructed. This model is built using both MATLAB/SIMULINK and FORTRAN languages. In this model, the full robot dynamics is used together with the realistic factors, such as the actuator effects and the gear backlash, to represent the real system accurately. All simulations throughout this thesis are carried out on this model. A model predictive control strategy for robot trajectory tracking is also introduced in this thesis. The feasibility of the proposed MPC control method is studied based on a perfect prediction model, a model with uncertainties, and when the frequency band of the MPC controller is limited. Furthermore, a new method of using neural networks for robot dynamics modelling is introduced. This method is developed on the basis of a numerical differential technique that eliminates the explicit requirement of robot joint accelerations. Therefore, this method can be easily implemented on physical systems. As the measurements of the robot joint positions, velocities, and torques collected from operating the robot can be used to train the neural network, a more accurate dynamic model can be obtained. Finally, the MPC control method and the neural network model are combined together to form a neural network based MPC controller. The validity of this method is verified by using simulation on the simulated robot system

Thesis (M.Sc. (Hons.)) -- University of Western Sydney, Nepean, 1999.
"A thesis submitted to the School of Mechatronic, Computer and Electrical Engineering, the University of Western Sydney, Nepean in fulfilment of the requirements for the degree of Master of Engineering (Honours)" Bibliography : leaves 119-123.

A trend in design and implementation of modern industrial automation systems is to integrate computing, communication and control into a unified framework at different levels of machine/factory operations and information processing. These distributed control systems are referred to as networked control systems (NCSs). They are composed of sensors, actuators, and controllers interconnected over communication networks. As most of communication networks are not designed for NCS applications, the communication requirements of NCSs may be not satisfied. For example, traditional control systems require the data to be accurate, timely and lossless. However, because of random transmission delays and packet losses, the control performance of a control system may be badly deteriorated, and the control system rendered unstable. The main challenge of NCS design is to both maintain and improve stable control performance of an NCS. To achieve this, communication and control methodologies have to be designed. In recent decades, Ethernet and 802.11 networks have been introduced in control networks and have even replaced traditional fieldbus productions in some real-time control applications, because of their high bandwidth and good interoperability. As Ethernet and 802.11 networks are not designed for distributed control applications, two aspects of NCS research need to be addressed to make these communication networks suitable for control systems in industrial environments. From the perspective of networking, communication protocols need to be designed to satisfy communication requirements for NCSs such as real-time communication and high-precision clock consistency requirements. From the perspective of control, methods to compensate for network-induced delays and packet losses are important for NCS design. To make Ethernet-based and 802.11 networks suitable for distributed control applications, this thesis develops a high-precision relative clock synchronisation protocol and an analytical model for analysing the real-time performance of 802.11 networks, and designs a new predictive compensation method. Firstly, a hybrid NCS simulation environment based on the NS-2 simulator is designed and implemented. Secondly, a high-precision relative clock synchronization protocol is designed and implemented. Thirdly, transmission delays in 802.11 networks for soft-real-time control applications are modeled by use of a Markov chain model in which real-time Quality-of- Service parameters are analysed under a periodic traffic pattern. By using a Markov chain model, we can accurately model the tradeoff between real-time performance and throughput performance. Furthermore, a cross-layer optimisation scheme, featuring application-layer flow rate adaptation, is designed to achieve the tradeoff between certain real-time and throughput performance characteristics in a typical NCS scenario with wireless local area network. Fourthly, as a co-design approach for both a network and a controller, a new predictive compensation method for variable delay and packet loss in NCSs is designed, where simultaneous end-to-end delays and packet losses during packet transmissions from sensors to actuators is tackled. The effectiveness of the proposed predictive compensation approach is demonstrated using our hybrid NCS simulation environment.

Indoor climate control is responsible for a substantial amount of the world's total energy expenditure. In a time of climate crisis where a reduction of energy consumption is crucial to avoid climate disaster, indoor climate control is a ripe target for eliminating energy waste. The conventional method of adjusting the indoor climate with the use of setpoint curves, based solely on outdoor temperature, may lead to notable inefficiencies. This project evaluates the possibility to replace this method of regulation with a system based on model predictive control (MPC) in one of Uppsala University Hospitals office buildings. A prototype of an MPC controller using Artificial Neural Networks (ANN) as its system model was developed. The system takes several data sources into account, including indoor and outdoor temperatures, radiator flowline and return temperatures, current solar radiation as well as forecast for both solar radiation and outdoor temperature. The system was not set in production but the controller's predicted values correspond well to the buildings current thermal behaviour and weather data. These theoretical results attest to the viability of using the method to regulate the indoor climate in buildings in place of setpoint curves.
Bibehållande av inomhusklimat står för en avsevärd del av världens totala energikonsumtion. Med dagens klimatförändringar där minskad energikonsumtion är viktigt för den hållbara utvecklingen så är inomhusklimat ett lämpligt mål för att förhindra slösad energi. Konventionell styrning av inomhusklimat använder sig av börvärdeskurvor, baserade enbart på utomhustemperatur, vilket kan leda till anmärkningsvärt energispill. Detta projekt utvärderar möjligheten att ersätta denna styrmetod med ett system baserat på model predictive control (MPC) och använda detta i en av Akademiska sjukhusets lokaler i Uppsala. En MPC styrenhet som använder Artificiella Neurala Nätverk (ANN) som sin modell utvecklades. Systemet använder sig av flera datakällor däribland inomhus- och utomhustemperatur, radiatorslingornas framlednings- och returtemperatur, rådande solinstrålning såväl som prognoser för solinstrålning och utomhustemperatur. Systemet sattes inte i produktion men dess prognos stämmer väl överens med tillgänglig väderdata och husets termiska beteende. De presenterade resultaten påvisar metoden vara ett lämpligt substitut för styrning med börvärdeskurvor.

Thesis (M. S.)--Aerospace Engineering, Georgia Institute of Technology, 2005.
Bottasso, Carlo, Committee Chair ; Hodges, Dewey, Committee Member ; Bauchau, Olivier, Committee Member. Includes bibliographical references.

Over the past decades, attention has been brought to the importance of indoor air quality and the serious threat of bio-aerosol contamination towards human health. A novel idea to transport hazardous particles away from sensitive areas is to automatically control bio-aerosol concentrations, by utilising airflows from ventilation systems. Regarding this, computational fluid dynamics (CFD) may be employed to investigate the dynamical behaviour of airborne particles, and data-driven methods may be used to estimate and control the complex flow simulations. This thesis presents a methodology for machine-learning based control of particle concentrations in turbulent gas-solid flow. The aim is to reduce concentration levels at a 90 degree corner, through systematic manipulation of underlying two-phase flow dynamics, where an energy constrained inlet airflow rate is used as control variable. A CFD experiment of turbulent gas-solid flow in a two-dimensional corner geometry is simulated using the SST k-omega turbulence model for the gas phase, and drag force based discrete random walk for the solid phase. Validation of the two-phase methodology is performed against a backwards facing step experiment, with a 12.2% error correspondence in maximum negative particle velocity downstream the step. Based on simulation data from the CFD experiment, a linear auto-regressive with exogenous inputs (ARX) model and a non-linear ARX based neural network (NN) is used to identify the temporal relationship between inlet flow rate and corner particle concentration. The results suggest that NN is the preferred approach for output predictions of the two-phase system, with roughly four times higher simulation accuracy compared to ARX. The identified NN model is used in a model predictive control (MPC) framework with linearisation in each time step. It is found that the output concentration can be minimised together with the input energy consumption, by means of tracking specified target trajectories. Control signals from NN-MPC also show good performance in controlling the full CFD model, with improved particle removal capabilities, compared to randomly generated signals. In terms of maximal reduction of particle concentration, the NN-MPC scheme is however outperformed by a manually constructed sine signal. In conclusion, CFD based NN-MPC is a feasible methodology for efficient reduction of particle concentrations in a corner area; particularly, a novel application for removal of indoor bio-aerosols is presented. More generally, the results show that NN-MPC may be a promising approach to turbulent multi-phase flow control.

As the greenhouse effect is an imminent concern, motivation for the development of energy efficient systems has grown fast. Today heavy-duty vehicles (HDVs) account for a growing part of the emissions from the vehicular transport sector. One way to reduce those emissions is by driving at short intervehicular distances in so called platoons, mainly on highways. In such formations, the aerodynamic drag is decreased which allows for more fuel efficient driving, meanwhile the roads are used more efficiently. This thesis deals with the question of how those platoons can be controlled without using communications between the involved HDVs. In this thesis, artificial neural networks are designed and trained to predict the velocity profile for an HDV driving over a section of road where data on the topography are available. This information is used in a model predictive controller to control the HDV driving behind the truck for which the aforementioned prediction is made. By having accurate information about the upcoming behaviour of the preceding HDV, the controller can plan the velocity profile for the controlled HDV in a way which minimizes fuel consumption. To ensure fuel optimal performance, a state describing the mass of consumed fuel is derived and minimized in the controller. A system modelling gear shift dynamics is proposed to capture essential dynamics such as torque loss during shifting. The designed controller is able to predict and change between the three highest gears making it able to handle almost all highway platooning scenarios. The prediction system shows great potential and is able to predict the velocity profile for different HDVs with an average error as low as 0.04 km/h. The controller is implemented in a simulation environment and results show that compared to a platoon without these predictions of the preceding HDV, the fuel consumption for the controlled HDV can be reduced by up to 6 %.

This thesis is devoted to design Model Predictive Control (MPC) strategies aiming to enhance the management of constrained generalised flow-based networks, with special attention to the economic optimisation and robust performance of such systems. Several control schemes are developed in this thesis to exploit the available economic information of the system operation and the disturbance information obtained from measurements and forecasting models. Dynamic network flows theory is used to develop control-oriented models that serve to design MPC controllers specialised for flow networks with additive disturbances and periodically time-varying dynamics and costs. The control strategies developed in this thesis can be classified in two categories: centralised MPC strategies and non-centralised MPC strategies. Such strategies are assessed through simulations of a real case study: the Barcelona drinking water network (DWN). Regarding the centralised strategies, different economic MPC formulations are first studied to guarantee recursive feasibility and stability under nominal periodic flow demands and possibly time-varying economic parameters and multi-objective cost functions. Additionally, reliability-based MPC, chance-constrained MPC and tree-based MPC strategies are proposed to address the reliability of both the flow storage and the flow transportation tasks in the network. Such strategies allow to satisfy a customer service level under future flow demand uncertainty and to efficiently distribute overall control effort under the presence of actuators degradation. Moreover, soft-control techniques such as artificial neural networks and fuzzy logic are used to incorporate self-tuning capabilities to an economic certainty-equivalent MPC controller. Since there are objections to the use of centralised controllers in large-scale networks, two non-centralised strategies are also proposed. First, a multi-layer distributed economic MPC strategy of low computational complexity is designed with a control topology structured in two layers. In a lower layer, a set of local MPC agents are in charge of controlling partitions of the overall network by exchanging limited information on shared resources and solving their local problems in a hierarchical-like fashion. Moreover, to counteract the loss of global economic information due to the decomposition of the overall control task, a coordination layer is designed to influence non-iteratively the decision of local controllers towards the improvement of the overall economic performance. Finally, a cooperative distributed economic MPC formulation based on a periodic terminal cost/region is proposed. Such strategy guarantees convergence to a Nash equilibrium without the need of a coordinator and relies on an iterative and global communication of local controllers, which optimise in parallel their control actions but using a centralised model of the network.
Esta tesis se enfoca en el diseño de estrategias de control predictivo basado en modelos (MPC, por sus siglas en inglés) con la meta de mejorar la gestión de sistemas que pueden ser descritos por redes generalizadas de flujo y que están sujetos a restricciones, enfatizando especialmente en la optimización económica y el desempeño robusto de tales sistemas. De esta manera, varios esquemas de control se desarrollan en esta tesis para explotar tanto la información económica disponible de la operación del sistema como la información de perturbaciones obtenida de datos medibles y de modelos de predicción. La teoría de redes dinámicas de flujo es utilizada en esta tesis para desarrollar modelos orientados a control que sirven para diseñar controladores MPC especializados para la gestión de redes de flujo que presentan tanto perturbaciones aditivas como dinámicas y costos periódicamente variables en el tiempo. Las estrategias de control propuestas en esta tesis se pueden clasificar en dos categorías: estrategias de control MPC centralizado y estrategias de control MPC no-centralizado. Dichas estrategias son evaluadas mediante simulaciones de un caso de estudio real: la red de transporte de agua potable de Barcelona en España. En cuanto a las estrategias de control MPC centralizado, diferentes formulaciones de controladores MPC económicos son primero estudiadas para garantizar factibilidad recursiva y estabilidad del sistema cuya operación responde a demandas nominales de flujo periódico, a parámetros económicos posiblemente variantes en el tiempo y a funciones de costo multi-objetivo. Adicionalmente, estrategias de control MPC basado en fiabilidad, MPC con restricciones probabilísticas y MPC basado en árboles de escenarios son propuestas para garantizar la fiabilidad tanto de tareas de almacenamiento como de transporte de flujo en la red. Tales estrategias permiten satisfacer un nivel de servicio al cliente bajo incertidumbre en la demanda futura, así como distribuir eficientemente el esfuerzo global de control bajo la presencia de degradación en los actuadores del sistema. Por otra parte, técnicas de computación suave como redes neuronales artificiales y lógica difusa se utilizan para incorporar capacidades de auto-sintonía en un controlador MPC económico de certeza-equivalente. Dado que hay objeciones al uso de control centralizado en redes de gran escala, dos estrategias de control no-centralizado son propuestas en esta tesis. Primero, un controlador MPC económico distribuido de baja complejidad computacional es diseñado con una topología estructurada en dos capas. En una capa inferior, un conjunto de controladores MPC locales se encargan de controlar particiones de la red mediante el intercambio de información limitada de los recursos físicos compartidos y resolviendo sus problemas locales de optimización de forma similar a una secuencia jerárquica de solución. Para contrarrestar la pérdida de información económica global que ocurra tras la descomposición de la tarea de control global, una capa de coordinación es diseñada para influenciar no-iterativamente la decisión de los controles locales con el fin de lograr una mejora global del desempeño económico. La segunda estrategia no-centralizada propuesta en esta tesis es una formulación de control MPC económico distribuido cooperativo basado en una restricción terminal periódica. Tal estrategia garantiza convergencia a un equilibrio de Nash sin la necesidad de una capa de coordinación pero requiere una comunicación iterativa de información global entre todos los controladores locales, los cuales optimizan en paralelo sus acciones de control utilizando un modelo centralizado de la red.

El control en temps real de xarxes de clavegueram (RTC) desenvolupa un paper fonamental dins de la gestió dels recursos hídrics relacionats amb el cicle urbà de l'aigua i, en general, amb el seu cicle natural. Un adequat disseny de control per a xarxes de clavegueram evita impactes mediambientals negatius originats per inundacions i/o alta pol·lució producte de condicions meteorològiques xtremes. No obstant, s'ha de tenir en compte que aquestes xarxes, a més de la seva grandària i quantitat de variables i instrumentació, són sistemes rics en dinàmiques complexes i altament no lineals. Aquest fet, unit a les condicions atmosfèriques extremes, fan necessari utilitzar una estratègia de control capaç¸ de suportar totes aquestes condicions. En aquest sentit, dins del camp del (RTC) de xarxes de clavegueram es destaquen les estratègies de control predictiu basat en model (MPC), les quals són alternatives adequades per al control de configuracions multivariable i de gran escala, aplicades com estratègies de control global del sistema. A m´es, permeten optimitzar la resposta del sistema tenint en compte diversos índexs de rendiment (control multiobjectiu).
Aquesta tesi s'enfoca en el disseny de controladors MPC per a xarxes de clavegueram considerant diverses metodologies de modelat. Addicionalment, analitza les situacions en les quals es presenten fallades als actuadors de la xarxa, proposant estratègies per a mantenir la resposta del sistema amb la menor degradació possible dels objectius de control, malgrat la presència de la fallada. En la primera part s'introdueixen els conceptes principals dels temes a tractar en la tesi: xarxes de clavegueram, MPC i tolerància a fallades. Seguidament, es presenta la tècnica de modelat utilitzada per a definir el model d'una xarxa de clavegueram. Finalment, es presenta i descriu el cas d'aplicació en la tesi: la xarxa de clavegueram de Barcelona (Espanya).
La segona part es centra en dissenyar controladors MPC per al cas d'estudi. S'han considerat dos tipus de model de xarxa: (i) un model lineal, el qual aproxima els comportaments no lineals de la xarxa, donant origen a estratègies MPC lineals amb les seves conegudes avantatges de l'optimització convexa i escalabilitat; i (ii) un model híbrid, el qual inclou les dinàmiques de commutació més representatives d'una xarxa de clavegueram com són els sobreeixidors.
En aquest últim cas es proposa una nova etodologia de modelat híbrid per a xarxes de clavegueram i es dissenyen estratègies de control predictives basades en aquests models (HMPC), les quals calculen lleis de control globalment òptimes. Addicionalment, es proposa una estratègia de relaxació del problema d'optimització discreta per a evitar els grans temps de còmput requerits per a calcular la llei de control HMPC.
Finalment, la tercera part de la tesi s'encarrega d'estudiar les capacitats de tolerància a fallades en actuadors de llaços de control MPC. En el cas de xarxes de clavegueram, la tesi considera fallades en les comportes de derivació i de retenció d'aigües residuals. A més, es proposa un modelat híbrid per a fallades que faci que el problema d'optimització associat no perdi la seva convexitat. Així, es proposen dos estratègies de HMPC tolerant a fallades (FTMPC): l'estratègia activa, la qual utilitza les avantatges d'una arquitectura de control tolerant a fallades (FTC), i l'estratègia passiva, la qual només depèn de la robustesa intrínseca de les tècniques de control MPC. Com a extensió a l'estudi de tolerància a fallades, es proposa una avaluació d'admissibilitat per a configuracions d'actuadors en fallada agafant com a referència la degradació dels objectius de control. El m-etode, basat en satisfacció de restriccions, permet avaluar l'admissibilitat d'una configuració d'actuadors en fallada i, en cas de no ser admesa, evitaria el procés de resoldre un problema d'optimització amb un alt cost computacional.
Paraules clau: control predictiu basat en model, sistemes de clavegueram, sistemes híbrids, MLD, control tolerant a fallades, satisfacció de restriccions.
El control en tiempo real de redes de alcantarillado (RTC) desempeña un papel fundamental dentro de la gestión de los recursos hídricos relacionados con el ciclo urbano del agua y, en general, con su ciclo natural. Un adecuado diseño de control para de redes de alcantarillado evita impactos medioambientales negativos originados por inundaciones y/o alta polución producto de condiciones meteorológicas extremas. Sin embargo, se debe tener en cuenta que estas redes, además de su gran tamaño y cantidad de variables e instrumentación, son sistemas ricos en dinámicas complejas y altamente no lineales. Este hecho, unido a unas condiciones atmosféricas extremas, hace necesario utilizar una estrategia de control capaz de soportar todas estas condiciones. En este sentido, dentro del campo del RTC de redes de alcantarillado se destacan las estrategias de control predictivo basadas en modelo (MPC), las cuales son alternativas adecuadas para el control de configuraciones multivariable y de gran escala, aplicadas como estrategias de control global del sistema. Además, permiten optimizar el desempeño del sistema teniendo en cuenta diversos índices de rendimiento (control multiobjetivo).
Esta tesis se enfoca en el diseño de controladores MPC para redes de alcantarillado considerando diversas metodologías de modelado. Adicionalmente, analiza las situaciones en las cuales se presentan fallos en los actuadores de la red, proponiendo estrategias para mantener el desempeño del sistema y evitando la degradación de los objetivos de control a pesar de la presencia del fallo. En la primera parte se introducen los conceptos principales de los temas a tratar en la tesis: redes de alcantarillado, MPC y tolerancia a fallos. Además, se presenta la técnica de modelado utilizada para definir el modelo de una red de alcantarillado. Finalmente, se presenta y describe el caso de aplicación considerado en la tesis: la red de alcantarillado de Barcelona (España).
La segunda parte se centra en diseñar controladores MPC para el caso de estudio. Dos tipos de modelo de la red son considerados: (i) un modelo lineal, el cual aproxima los comportamientos no lineales de la red, dando origen a estrategias MPC lineales con sus conocidas ventajas de optimización convexa y escalabilidad; y (ii) un modelo híbrido, el cual incluye las dinámicas de conmutación más representativas de una red de alcantarillado como lo son los rebosaderos. En este último caso se propone una nueva metodología de modelado híbrido para redes de alcantarillado y se diseñan estrategias de control predictivas basadas en estos modelos (HMPC), las cuales calculan leyes de control globalmente óptimas. Adicionalmente se propone una estrategia de relajación del problema de optimización discreto para evitar los grandes tiempos de cálculo que pudieran ser requeridos al obtener la ley de control HMPC.
Finalmente, la tercera parte de la tesis se ocupa de estudiar las capacidades de tolerancia a fallos en actuadores de lazos de control MPC. En el caso de redes de alcantarillado, la tesis considera fallos en las compuertas de derivación y de retención de aguas residuales. De igual manera, se propone un modelado híbrido para los fallos que haga que el problema de optimización asociado no pierda su convexidad. Así, se proponen dos estrategias de HMPC tolerante a fallos (FTMPC): la estrategia activa, la cual utiliza las ventajas de una arquitectura de control tolerante a fallos (FTC), y la estrategia pasiva, la cual sólo depende de la robustez intrínseca de las técnicas de control MPC. Como extensión al estudio de tolerancia a fallos, se propone una evaluación de admisibilidad para configuraciones de actuadores en fallo tomando como referencia la degradación de los objetivos de control. El método, basado en satisfacción de restricciones, permite evaluar la admisibilidad de una configuración de actuadores en fallo y, en caso de no ser admitida, evitaría el proceso de resolver un problema de optimización con un alto coste computacional.
Palabras clave: control predictivo basado en modelo, sistemas de alcantarillado, sistemas híbridos, MLD, control tolerante a fallos, satisfacción de restricciones.
Real time control (RTC) of sewer networks plays a fundamental role in the management of hydrological systems, both in the urban water cycle, as well as in the natural water cycle. An adequate design of control systems for sewer networks can prevent the negative impact on the environment that Combined Sewer Overflow (CSO) as well as preventing flooding within city limits when extreme weather conditions occur. However, sewer networks are large scale systems with many variables, complex dynamics and strong nonlinear behaviour. Any control strategy applied should be capable of handling these challenging requirements. Within the field of RTC of sewer networks for global network control, the Model Predictive Control (MPC) strategy stands out due to its ability to handle large scale, nonlinear and multivariable systems. Furthermore, this strategy allows performance optimization, taking into account several control objectives simultaneously.
This thesis is devoted to the design of MPC controllers for sewer networks, as well as the complementary modelling methodologies. Furthermore, scenarios where actuator faults occur are specially considered and strategies to maintain performance or at least minimizing its degradation in presence of faults are proposed. In the first part of this thesis, the basic concepts are introduced: sewer networks, MPC and fault tolerant control. In addition, the modelling methodologies used to describe such systems are presented. Finally the case study of this thesis is described: the sewer network of the city of Barcelona (Spain). The second part of this thesis is centered on the design of MPC controllers for the proposed case study. Two types of models are considered: (i) a linear model whose corresponding MPC strategy is known for its advantages such as convexity of the optimization problem and existing pro of sofstability, and (ii) a hybrid model which allows the inclusion of state dependent hybrid dynamics such as weirs. In the latter case, a new hybrid modelling methodology is introduced and hybrid model predictive control (HMPC) strategies based on these models are designed. Furthermore, strategies to relax the optimization problem are introduced to reduce calculation time required for the HMPC control law.
Finally, the third part of this thesis is devoted to study the fault tolerance capabilities of MPC controllers. Actuator faults in retention and redirection gates are considered. Additionally, hybrid modelling techniques are presented for faults which, in the linear case, can not be treated without loosing convexity of the related optimization problem. Two fault tolerant HMPC strategies are compared: the active strategy, which uses the information from a diagnosis system to maintain control performance, and the passive strategy which only relies on the intrinsic robustness of the MPC control law. As an extension to the study of fault tolerance, the admissibility of faulty actuator configurations is analyzed with regard to the degradation of control objectives. The method, which is based on constraint satisfaction, allows the admissibility evaluation of actuator fault configurations, which avoids the process of solving the optimization problem with its related high computational cost.
Keywords: MPC, sewer networks, hybrid systems, MLD, fault tolerant control, constraints satisfaction.

Model predictive control (MPC) is a paradigm within automatic control notable for its ability to handle constraints. This ability come at the cost of high computational demand, which until recently has limited use of MPC to slow systems. Recent advances have however enabled MPC to be used in embedded applications, where its ability to handle constraints can be leveraged to reduce wear, increase efficiency and improve overall performance in everything from cars to wind turbines. MPC controllers can be made even faster by precomputing the resulting policy and storing it in a lookup table. A method known as explicit MPC. An alternative way of leveraging precomputation is to train a neural network to approximate the policy. This is an attractive proposal both due to neural networks ability to imitate policies for nonlinear systems, and results that indicate that neural networks can efficiently represent explicit MPC policies. Limited work has been done in this area. How the networks are setup and trained therefore tends to reflect recent trends in other application areas rather than being based on what is known to work well for approximating MPC policies. This thesis attempts to alleviate this situation by evaluating how some common neural network architectures and training methods performs when used for this purpose. The evaluations are carried out through a literature study and by training several networks with different architectures to replicate the policy of a nonlinear MPC controller tasked with stabilizing an inverted pendulum. The results suggest that ReLU activation functions give better performance than hyperbolic tangent and SELU functions; and that dropout and batch normalization degrades the ability to approximate policies; and that depth significantly increases the performance. However, the neural network controllers do occasionally exhibit problematic behaviors, such as steady state errors and oscillating control signals close to constraints.
Modell-prediktiv reglering (MPC, efter engelskans Model Predictive Control) är ett paradigm inom reglertekniken som på ett effektivt sätt kan hantera begränsningar i systemet som ska regleras. Den här egenskapen kommer på bekostnad av att MPC kräver mycket datorkraft. Tidigare har  användning av den här typen av kontroller därför varit begränsad till långsamma system. På senare tid har framsteg inom hård- och mjukvara dock möjliggjort användning av MPC på inbyggda system. Där kan dess förmåga att hantera begränsningar användas för att minska slitage, öka effektivitet och förbättra prestanda inom allt från bilar till vindkraftverk. Ett sätt att minska beräkningsbördan ytterligare är att beräkna MPC-policyn i förväg och spara den i en tabell. Det här tillvägagångssättet kallas explicit MPC. Ett alternativt tillvägagångssätt är att träna ett neuralt nätverk till att approximera policyn. Potentiellt har det här fördelarna att ett neuralt nätverk inte är begränsat till att efterlikna policys för system med linjär dynamik, och att det finns resultat som pekar på att neurala nätverk är väl lämpade för att lagra policys för explicit MPC. En begränsad mängd arbete har gjorts inom det här området. Hur nätverken designas och tränas tenderar därför att reflektera trender inom andra applikationsområden för neurala nätverk istället för att baseras på vad som fungerar för att implementera MPC. Det här examensarbetet försöker avhjälpa det här problemet. Dels genom en litteraturstudie och dels genom att undersöka hur olika arkitekturer för neurala nätverk beter sig när de tränas för att efterlikna en ickelinjär MPC-kontroller som ska stabilisera en inverterad pendel. Resultaten tyder på att nätverk med ReLU-aktivering ger bättre prestanda än motsvarande nätverk som använder SELU eller tangens hyperbolicus som aktiveringsfunktion. Resultaten visar också att batch noralization och dropout försämmrar nätverkens förmåga att lära sig policyn och att prestandan blir bättre om antalet lager i nätverket ökar. De neurala nätverken uppvisar dock i vissa fall kvalitativa problem, så som statiska fel och oscillerande kontrollsignaler nära begränsningar.

Growth in Distributed Energy Resources (DERs) and low-inertia renewable energy sources in smart grids require imperative Volt-VAR Control (VVC). Moreover, this growth combined with increasing deployment of information technologies in smart grids fuels communication uncertainties and reveals transient stability challenges for Distributed Network Operators (DNOs). Innovative approaches have been proposed to use the inherent thermal inertia of buildings to provide ancillary services to the grid to tackle the problems posed by the increasing trend of volatile DERs. Although numerous approaches harness traditional VVC devices to compensate for voltage violations, synthetic inertia and control of Energy Storage System (ESS) exist to improve transient stability with an increase of DERs. While ample strategies tackle these two problems separately, the ability of smart buildings to provide active and reactive power support simultaneously has not yet been exploited. This study explores the concurrent effects of modulating loads’ apparent power consumption on the grid’s frequency and voltage profile. A Distributed Model Predictive Control (DMPC) strategy for voltage and frequency control in the DN is employed by using smart buildings and sensitivity analysis without compromising customers’ climate control performance in smart buildings. The robustness of this strategy is validated on a modified IEEE 13 bus system modelled in MathWorks Simulink.

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.

Diabetes is a group of metabolic diseases where the body’s pancreas does not produce enough insulin or does not properly respond to insulin produced, resulting in high blood sugar levels over a prolonged period. There are several different types of diabetes, but the most common forms are type 1 and type 2 diabetes. Type 1 diabetes Mellitus (T1DM) can occur at any age, but is most commonly diagnosed from infancy to late 30s. If a person is diagnosed with type 1 diabetes, their pancreas produces little to no insulin, and the body’s immune system destroys the insulin-producing cells in the pancreas. Those diagnosed with type 1 diabetes must inject insulin several times every day or continually infuse insulin through a pump, as well as manage their diet and exercise habits. If not treated appropriately, it can cause serious complications such as cardiovascular disease, stroke, kidney failure, foot ulcers, and damage to eyes.

During the past decade, researchers have developed artificial pancreas (AP) to ease management of diabetes. AP has three components: continuous glucose monitor (CGM), insulin pump, and closed-loop control algorithm. Researchers have developed algorithms based on control techniques such as Proportional Integral Derivative (PID) and Model Predictive Control (MPC) for blood glucose level (BGL) control; however, variability in metabolism between or within individuals hinders reliable control.

This study aims to develop an adaptive algorithm using Artificial Neural Networks (ANN) based Model Predictive Control (NN-MPC) to perform proper insulin injections according to BGL predictions in diabetic rats. This study is a ground work to implement NN-MPC algorithm on real subjects. BGL data collected from diabetic rats using CGM are used with other inputs such as insulin injection and meal information to develop a virtual plant model based on a mathematical model of glucose–insulin homeostasis proposed by Lombarte et al. Since this model is proposed for healthy rats; a revised version on this model with three additional equations representing diabetic rats is used to generate data for training ANN which is applicable for the identi?cation of dynamics and the glycemic regulation of rats. The trained ANN is coupled with MPC algorithm to control BGL of the plant model within the normal range of 100 to 130 mg/dl by injecting appropriate amount of insulin. The ANN performed well with less than 5 mg/dl error (2%) for 5-minute prediction and about 15 mg/dl error (7%) for 30-minute prediction. In ¬¬addition, the NN-MPC algorithm kept BGL of diabetic rats more than 90 percent of the time within the normal range without hyper/hypo-glycaemia.

Submitted by Felipe Fakir null (zafakir@yahoo.com.br) on 2017-06-27T07:01:28Z No. of bitstreams: 1 FFAKIR Dissertação vFinalFichaCataAta.pdf: 2064786 bytes, checksum: 158a935a636b9dbf9e59618a35b4c8ef (MD5)
Approved for entry into archive by Luiz Galeffi (luizgaleffi@gmail.com) on 2017-06-28T19:39:58Z (GMT) No. of bitstreams: 1 fakir_f_me_bauru.pdf: 2064786 bytes, checksum: 158a935a636b9dbf9e59618a35b4c8ef (MD5)
Made available in DSpace on 2017-06-28T19:39:58Z (GMT). No. of bitstreams: 1 fakir_f_me_bauru.pdf: 2064786 bytes, checksum: 158a935a636b9dbf9e59618a35b4c8ef (MD5) Previous issue date: 2017-06-01
A tecnologia de comunicação wireless vem se tornando parte fundamental do cotidiano das indústrias de processos, onde o uso de transmissores wireless aplicados à monitoração e controle já é uma realidade. A arquitetura de Sistema de Controle via Rede Sem Fio (WNCS) possui vantagens em relação às arquiteturas tradicionais ponto-a-ponto e às arquiteturas de redes cabeadas devido à facilidade de instalação, configuração e manutenção. No entanto, a evolução desta tecnologia introduziu novos desafios para a implementação da malha de controle fechada por um instrumento wireless como as não linearidades, perda de pacote de dados e restrições da comunicação de dados nas redes sem fio. Outro fator crítico relacionado à implementação de WNCSs é a fonte de energia limitada destes transmissores, que possuem vida útil dependente da quantidade de acessos e dados transmitidos. Este trabalho apresenta o estudo e o desenvolvimento de um controlador preditivo multi-rate como alternativa para melhorar a eficiência energética em aplicações industriais de WNCSs. A estratégia proposta não necessita receber constantemente os valores reais das variáveis do processo transmitidos pelos transmissores wireless, pois o controlador preditivo baseado em modelo (MPC) se utiliza do submodelo interno das variáveis de processo para estimar os valores das variáveis quando estas não são transmitidas. Dessa forma, uma diminuição da frequência de transmissão de dados na rede sem fio pode ser obtida e, consequentemente uma redução do consumo energético dos dispositivos sem fio. Resultados de simulações em diferentes condições de operação de um WNCS multivariável de controle de tanques acoplados demonstram que o MPC multi-rate possui características de robustez e é efetivo para aplicações de WNCS, garantindo requisitos de controle e estabilidade mesmo com a diminuição da frequência de transmissão de dados de realimentação na rede sem fio. Adicionalmente, resultados do consumo energético dos dispositivos do WNCS mostraram que o MPC multi-rate proporciona uma economia de energia de até 20% das baterias dos transmissores wireless. Uma análise da eficiência energética do WNCS é apresentada através do estudo dos limites operacionais do controlador MPC multi-rate considerando a relação de compromisso entre o período de amostragem dos dispositivos sem fio e o desempenho de controle do WNCS.
Wireless communication technology has become a fundamental part of the everyday life of process industries, where the use of wireless transmitters for monitoring and control is already a reality. The architecture of Wireless Networked Control Systems (WNCSs) has advantages over point-to-point and wired networks architectures due to the ease of installation, configuration and maintenance. However, the evolution of this technology has introduced new challenges to the implementation of the closed loop control with a wireless instrument as nonlinearities, packet losses and data communication constraints in the wireless networks. Another critical factor related to implementation of WNCSs is the energy source of these transmitters, which have limited lifetime dependent on the amount of access and data transmitted. This work presents the study and the development of a multi-rate predictive controller as an alternative to improve energy efficiency in industrial applications of WNCSs. The proposed strategy does not need to frequently receive updated process variables transmitted by wireless transmitters, because the model predictive controller (MPC) uses the internal submodel of the process variables to estimate the variables values when they are not transmitted. Thus, a decrease in the frequency of data transmission on the wireless network can be obtained and consequently a reduction of energy consumption of wireless devices. Simulation results for different operating conditions of a multivariable WNCS of coupled tanks shows that the multi-rate MPC provides robustness and it is effective for WNCS applications, ensuring control and stability requirements even with the reduction of the transmission frequency of the feedback data in the wireless network. In addition, energy consumption results from the WNCS devices showed that MPC multi-rate provides 20% of energy economy as it is effective in saving the energy expenditure of the wireless transmitter’s battery. An energy efficiency analysis of the WNCS is presented by studying the operating limits of the multi-rate MPC controller considering the compromise relationship between the sampling period of the wireless devices and the control performance of the WNCS.

In relation to the increasing interest in implementing smart cities, deployment of widespread wireless sensor networks (WSNs) has become a current hot topic. Among the application’s greatest challenges, there is still progress to be made concerning energy consumption and quality of service. Consequently, this project aims to explore a series of feasible solutions to improve the WSN energy efficiency for data aggregation by the WSN. This by strategically adjusting the position of the receiving base station and the packet rate of the WSN nodes. Additionally, the low-energy adaptive clustering hierarchy (LEACH) protocol is coupled with the WSN state of charge (SoC). For this thesis, a WSN was defined as a two dimensional area which contains sensor nodes and a mobile sink, i.e. a movable base station. Subsequent to the rigorous analyses of the WSN data clustering principles and system-wide dynamics, two different developing strategies, model-based and data-driven designs, were employed to develop two corresponding control approaches, model predictive control and reinforcement learning, on WSN energy management. To test their performance, a simulation environment was thus developed in Python, including the extended LEACH protocol. The amount of data transmitted by an energy unit is adopted as the index to estimate the control performance. The simulation results show that the model based controller was able to aggregate over 22% more bits than only using the LEACH protocol. Whilst the data driven controller had a worse performance than the LEACH network but showed potential for smaller sized WSNs containing a fewer amount of nodes. Nonetheless, the extension of the LEACH protocol did not give rise to obvious improvement on energy efficiency due to a wide range of differing results.
I samband med det ökande intresset för att implementera så kallade smart cities, har användningen av utbredda trådlösa sensor nätverk (WSN) blivit ett intresseområde. Bland applikationens största utmaningar, finns det fortfarande förbättringar med avseende på energiförbrukning och servicekvalité. Därmed så inriktar sig detta projekt på att utforska en mängd möjliga lösningar för att förbättra energieffektiviteten för dataaggregation inom WSN. Detta gjordes genom att strategiskt justera positionen av den mottagande basstationen samt paketfrekvensen för varje nod. Dessutom påbyggdes low-energy adaptive clustering hierarchy (LEACH) protokollet med WSN:ets laddningstillstånd. För detta examensarbete definierades ett WSN som ett två dimensionellt plan som innehåller sensor noder och en mobil basstation, d.v.s. en basstation som går att flytta. Efter rigorös analys av klustringsmetoder samt dynamiken av ett WSN, utvecklades två kontrollmetoder som bygger på olika kontrollstrategier. Dessa var en modelbaserad MPC kontroller och en datadriven reinforcement learning kontroller som implementerades för att förbättra energieffektiviteten i WSN. För att testa prestandan på dom två kontrollmetoderna, utvecklades en simulations platform baserat på Python, tillsamans med påbyggnaden av LEACH protokollet. Mängden data skickat per energienhet användes som index för att approximera kontrollprestandan. Simuleringsresultaten visar att den modellbaserade kontrollern kunde öka antalet skickade datapacket med 22% jämfört med när LEACH protokollet användes. Medans den datadrivna kontrollern hade en sämre prestanda jämfört med när enbart LEACH protokollet användes men den visade potential för WSN med en mindre storlek. Påbyggnaden av LEACH protokollet gav ingen tydlig ökning med avseende på energieffektiviteten p.g.a. en mängd avvikande resultat.

This work presents a model predictive control system for heating and cooling supply planning in an urban heating and cooling network. The control approach addresses the need for strategic operation of distributed production technologies and thermal energy storage in increasingly complex heating and cooling networks. Predictive optimization handles this complexity with an optimization strategy taking future demand, prices, and energy source availability into consideration. The model predictive control is integrated in a model built in a co-simulation approach. The co-simulation approach allows for models to run in their own simulation environments, preserving their levels of detail.  The model is adapted to a case study of an urban district under construction in Stockholm. Yearly simulations of the network and comparisons of the outcome when operated by the model predictive controller and by a reference rule-based controller are performed. The results show performance improvements in the form of reduced operational costs of 9.7 % and 18.8 % reduced carbon emissions, depending on how the objective function of the model predictive controller is formulated. An objective function aiming to minimize district heating imports is also formulated. While that objective function decreases the imports compared to the other objective functions, it increases the imports compared with the reference scenario, albeit from an already low share in the total energy supply of 0.2 %. A sensitivity analysis is performed to investigate the robustness of the control system. The sensitivity analysis shows that the reference controller is not robustly programmed for variations in parameters compared with the model predictive controller, which performs consistently better with both increases and decreases of the parameter sizes.  Future work could include detailed modelling in other simulation tools integrated in the co-simulation platform. Another possibility is developing a closed-loop system approach which would include, for example, feedback from the buildings’ indoor temperatures. This would allow for the utilisation of the buildings’ thermal mass as thermal energy storage. Lastly, more detailed economic and environmental calculations, such as life-cycle analysis or investment calculations, would further emphasize the real-world applicability of the findings.
Det här arbetet presenterar ett Model Predictive Control system för planering av värme- och kyltillförsel i ett urbant värme- och kylnät. Distribuerade energiresurser och termisk värmelagring leder till ökad komplexitet i planering och drift av framtidens värme- och kylnät. Prediktiv optimering hanterar komplexitet med en optimeringsstrategi som tar hänsyn till framtida efterfrågan, priser och tillgänglighet av energiresurser. Model Predictive Control systemet är integrerat i en modell uppbyggt i en Co-Simulation miljö. Co-Simulation möjliggör detaljerad modellering av olika delsystem i dess specifika simulerings miljö för att bevara dess detaljnivå.  Modellen är anpassad till en fallstudie av ett urbant distrikt under uppbyggnad i norra Stockholm. Årliga simuleringar av distriktet genomfördes. Därefter jämfördes resultat mellan simuleringar med Model Predictive Control systemet med ett konventionellt regel-baserat kontrollsystem. Tre målfunktioner var formulerade för Model Predictive Control systemet. Den första att minska driftkostnader för systemet, den andra att minska koldioxidutsläpp och det sista att minska importen från fjärrvärmenätet. Den första målfunktionen ger en minskning på 9.7 % i driftkostnader, den andra ger minskade koldioxidutsläpp på 18.8 %. Den tredje och sista däremot uppnår inte målet och ökar importen från fjärrvärmesystem jämfört med det konventionella regel-baserade kontrollsystemet. Utöver det så är en känslighetsanalys genomförd för att visa på robusthet av kontrollsystemen. Den visar att det Model Predictive Control systemet anpassar sig till förändringar i parametrar bättre än det andra kontrollsystemet.  Framtida arbeten inom området kan inkludera mer detaljerade modellering av de olika teknologierna inkluderade i studien. En annan möjlighet är utveckling av ett återkopplingssystem från byggnadernas inomhustemperatur. Det skulle möjliggöra användningen av byggnadens termiska massa som termisk energilagringssystem. Slutligen, mer detaljerad ekonomiska beräkningar och miljöberäkningar, såsom life-cycle analysis eller investeringskalkylering skulle utveckla resultaten från arbetet också.

Simulation of maneuvers with multibody models of rotorcraft vehicles is an important research area due to its complexity. During the maneuvering flight, some important design limitations are encountered such as maximum loads and maximum turning rates near the proximity of the flight envelope. This increases the demand on high fidelity models in order to define appropriate controls to steer the model close to the desired trajectory while staying inside the boundaries. A framework based on the hierarchical decomposition of the problem is used for this study. The system should be capable of generating the track by itself based on the given criteria and also capable of piloting the model of the vehicle along this track. The generated track must be compatible with the dynamic characteristics of the vehicle. Defining the constraints for the maneuver is of crucial importance when the vehicle is operating close to its performance boundaries. In order to make the problem computationally feasible, two models of the same vehicle are used where the reduced model captures the coarse level flight dynamics, while the fine scale comprehensive model represents the plant. The problem is defined by introducing planning layer and control layer strategies. The planning layer stands for solving the optimal control problem for a specific maneuver of a reduced vehicle model. The control layer takes the resulting optimal trajectory as an optimal reference path, then tracks it by using a non-linear model predictive formulation and accordingly steers the multibody model. Reduced models for the planning and tracking layers are adapted by using neural network approach online to optimize the predictive capabilities of planner and tracker. Optimal neural network architecture is obtained to augment the reduced model in the best way. The methodology of adaptive learning rate is experimented with different strategies. Some useful training modes and algorithms are proposed for these type of applications. It is observed that the neural network increased the predictive capabilities of the reduced model in a robust way. The proposed framework is demonstrated on a maneuvering problem by studying an obstacle avoidance example with violent pull-up and pull-down.

Quality of Service (QoS) in real-time media applications can be defined as the ability to guarantee the delivery of packets from source to destination over best-effort networks within some constraints. These constraints defined as the QoS metrics are end-to-end packet delay, delay jitter, throughtput, and packet losses. Transporting real-time media applications over best-effort networks, e.g. the Internet, is an area of current research. Both the Transmission Control Protocol (TCP) and the User Datagram Protocol (UDP) have failed to provide the desired QoS. This research aims at developing application-level end-to-end QoS controls to improve the user-perceived quality of real-time media applications over best-effort networks, such as, the public Internet. In this research an end-to-end packet based approach is developed. The end-to- end packet based approach consists of source buffer, network simulator ns-2, destina- tion buffer, and controller. Unconstrained model predictive control (MPC) methods are implemented by the controller at the application layer. The end-to-end packet based approach uses end-to-end network measurements and predictions as feedback signals. Effectiveness of the developed control methods are examined using Matlab and ns-2. The results demonstrate that sender-based control schemes utilizing UDP at transport layer are effective in providing QoS for real-time media applications transported over best-effort networks. Significant improvements in providing QoS are visible by the reduction of packet losses and the elimination of disruptions during the playback of real-time media. This is accompanied by either a decrease or increase in the playback start-time.

This thesis presents the development of self adaptive flight controller for an unmanned helicopter system under hovering manoeuvre. The neural network (NN) based model predictive control (MPC) approach is utilised in this work. We use this controller due to its ability to handle system constraints and the time varying nature of the helicopter dynamics. The non-linear NN based MPC controller is known to produce slow solution convergence due to high computation demand in the optimisation process. To solve this problem, the automatic flight controller system is designed using the NN based approximate predictive control (NNAPC) approach that relies on extraction of linear models from the non-linear NN model at each time step. The sequence of control input is generated using the prediction from the linearised model and the optimisation routine of MPC subject to the imposed hard constraints. In this project, the optimisation of the MPC objective criterion is implemented using simple and fast computation of the Hildreth's Quadratic Programming (QP) procedure. The system identification of the helicopter dynamics is typically performed using the time regression network (NNARX) with the input variables. Their time lags are fed into a static feed-forward network such as the multi-layered perceptron (MLP) network. NN based modelling that uses the NNARX structure to represent a dynamical system usually requires a priori knowledge about the model order of the system. Low model order assumption generally leads to deterioration of model prediction accuracy. Furthermore, massive amount of weights in the standard NNARX model can result in an increased NN training time and limit the application of the NNARX model in a real-time application. In this thesis, three types of NN architectures are considered to represent the time regression network: the multi-layered perceptron (MLP), the hybrid multi-layered perceptron (HMLP) and the modified Elman network. The latter two architectures are introduced to improve the training time and the convergence rate of the NN model. The model structures for the proposed architecture are selected using the proposed Lipschitz coefficient and k-cross validation methods to determine the best network configuration that guarantees good generalisation performance for model prediction. Most NN based modelling techniques attempt to model the time varying dynamics of a helicopter system using the off-line modelling approach which are incapable of representing the entire operating points of the flight envelope very well. Past research works attempt to update the NN model during flight using the mini-batch Levenberg-Marquardt (LM) training. However, due to the limited processing power available in the real-time processor, such approaches can only be employed to relatively small networks and they are limited to model uncoupled helicopter dynamics. In order to accommodate the time-varying properties of helicopter dynamics which change frequently during flight, a recursive Gauss-Newton (rGN) algorithm is developed to properly track the dynamics of the system under consideration. It is found that the predicted response from the off-line trained neural network model is suitable for modelling the UAS helicopter dynamics correctly. The model structure of the MLP network can be identified correctly using the proposed validation methods. Further comparison with model structure selection from previous studies shows that the identified model structure using the proposed validation methods offers improvements in terms of generalisation error. Moreover, the minimum number of neurons to be included in the model can be easily determined using the proposed cross validation method. The HMLP and modified Elman networks are proposed in this work to reduce the total number of weights used in the standard MLP network. Reduction in the total number of weights in the network structure contributes significantly to the reduction in the computation time needed to train the NN model. Based on the validation test results, the model structure of the HMLP and modified Elman networks are found to be much smaller than the standard MLP network. Although the total number of weights for both of the HMLP and modified Elman networks are lower than the MLP network, the prediction performance of both of the NN models are on par with the prediction quality of the MLP network. The identification results further indicate that the rGN algorithm is more adaptive to the changes in dynamic properties, although the generalisation error of repeated rGN is slightly higher than the off-line LM method. The rGN method is found capable of producing satisfactory prediction accuracy even though the model structure is not accurately defined. The recursive method presented here in this work is suitable to model the UAS helicopter in real time within the control sampling time and computational resource constraints. Moreover, the implementation of proposed network architectures such as the HMLP and modified Elman networks is found to improve the learning rate of NN prediction. These positive findings inspire the implementation of the real time recursive learning of NN models for the proposed MPC controller. The proposed system identification and hovering control of the unmanned helicopter system are validated in a 6 degree of freedom (DOF) safety test rig. The experimental results confirm the effectiveness and the robustness of the proposed controller under disturbances and parameter changes of the dynamic system.

The energy crisis together with greater environmental awareness, has increased interest in the construction of low energy buildings. Fabric thermal storage systems provide a promising approach for reducing building energy use and cost, and consequently, the emission of environmental pollutants. Hollow core ventilated slab systems are a form of fabric thermal storage system that, through the coupling of the ventilation air with the mass of the slab, are effective in utilizing the building fabric as a thermal store. However, the benefit of such systems can only be realized through the effective control of the thermal storage. This thesis investigates an optimum control strategy for the hollow core ventilated slab systems, that reduces the energy cost of the system without prejudicing the building occupants thermal comfort. The controller uses the predicted ambient temperature and solar radiation, together with a model of the building, to predict the energy costs of the system and the thermal comfort conditions in the occupied space. The optimum control strategy is identified by exercising the model with a numerical optimization method, such that the energy costs are minimized without violating the building occupant's thermal comfort. The thesis describes the use of an Auto Regressive Moving Average model to predict the ambient conditions for the next 24 hours. A building dynamic lumped parameter thermal network model, is also described, together with its validation. The implementation of a Genetic Algorithm search method for optimizing the control strategy is described, and its performance in finding an optimum solution analysed. The characteristics of the optimum schedule of control setpoints are investigated for each season, from which a simplified time-stage control strategy is derived. The effects of weather prediction errors on the optimum control strategy are investigated and the performance of the optimum controller is analysed and compared to a conventional rule-based control strategy. The on-line implementation of the optimal predictive controller would require the accurate estimation of parameters for modelling the building, which could form part of future work.

This thesis regards the problem of optimal management of water resources in inland navigation networks from a control theory perspective. In particular, the main objective to be attained consists in guaranteeing the navigability condition of the network, i.e., ensuring that the water levels are such that vessels can travel safely. More specifically, the water levels must be kept within an interval around the setpoint. Other common objectives include minimizing the operational cost and ensuring a long lifespan of the equipment. However, inland navigation networks are large-scale systems characterized by a number of features that complicate their management, namely complex dynamics, large time delays and negligible bottom slopes. In order to achieve the optimal management, the efficient control of the hydraulic structures, e.g., gates, weirs and locks, must be ensured. To this end, a control-oriented modeling approach is derived based on an existing simplified model obtained from the Saint-Venant equations. This representation reduces the complexity of the original model, provides flexibility and allows to coordinate current and delayed information in a systematic manner. However, the resulting model formulation belongs to the class of delayed descriptor systems, for which standard control and state estimation tools would need to be extended. Instead, model predictive control and moving horizon estimation can be easily adapted for this formulation, as well as being able to deal with physical and operational constraints in a natural manner. Due to the large dimensionality of inland navigation networks, a centralized implementation is often neither possible nor desirable. In this regard, non-centralized approaches are considered, decomposing the overall system in subsystems and distributing the computational burden among the local agents, each of them in charge of meeting the local objectives. Given the fact that inland navigation networks are strongly coupled systems, a distributed approach is followed, featuring a communication protocol among local agents. Despite the optimality of the computed solutions, state estimation will only be effective provided that the sensors acquire reliable data. Likewise, the control actions will only be applied correctly if the actuators are not impacted by faults. Indeed, any error can lead to an inefficient management of the system. Therefore, the last part of the thesis is concerned with the design of supervisory strategies that allow to detect and isolate faults in inland navigation networks. All the presented modeling, centralized and distributed control and state estimation and fault diagnosis approaches are applied to a realistic case study based on the inland navigation network in the north of France to validate their effectiveness.
Cette thèse contribue à répondre au problème de la gestion optimale des ressources en eau dans les réseaux de navigation intérieure du point de vue de la théorie du contrôle. Les objectifs principales à atteindre consistent à garantir la navigabilité des réseaux de voies navigables, veiller à la réduction des coûts opérationnels et à la longue durée de vie des équipements. Lors de la conception de lois de contrôle, les caractéristiques des réseaux doivent être prises en compte, à savoir leurs dynamiques complexes, des retards variables et l’absence de pente. Afin de réaliser la gestion optimale, le contrôle efficace des structures hydrauliques doit être assuré. A cette fin, une approche de modélisation orientée contrôle est dérivée. Cependant, la formulation obtenue appartient à la classe des systèmes de descripteurs retardés, pour lesquels la commande prédictive MPC et l’estimation d’état sur horizon glissant MHE peuvent être facilement adaptés à cette formulation, tout en permettant de gérer les contraintes physiques et opérationnelles de manière naturelle. En raison de leur grande dimensionnalité, une mise en œuvre centralisée n’est souvent ni possible ni souhaitable. Compte tenu du fait que les réseaux de navigation intérieure sont des systèmes fortement couplés, une approche distribuée est proposée, incluant un protocole de communication entre agents. Malgré l’optimalité des solutions, toute erreur peut entraîner une gestion inefficace du système. Par conséquent, les dernières contributions de la thèse concernent la conception de stratégies de supervision permettant de détecter et d’isoler les pannes des équipements. Toutes les approches présentées sont appliquées à une étude de cas réaliste basée sur le réseau de voies navigables du nord e la France afin de valider leur efficacité.
La present tesi versa sobre el problema de la gestió òptima dels recursos hídrics en vies de navegació interior des de la perspectiva de la teoria de control. Concretament, l’objectiu principal radica en garantir la condició de navegabilitat del s is tema. Dit d’una altra manera, es vol garantir que els nivells d’aigua siguin tals que les embarcacions puguin navegar-hi de forma segura. Aquest objectiu s’assoleix mantenint els nivells a l’interior d’un interval construït al voltant del punt d’operació. Altres objectius comuns en aquest context as piren a minimitzar els cos tos associats a l’operació dels equips, així com a prolongar-ne la seva vida útil. Ara bé, les vies de navegació interior són sistemes a gran escala caracteritzats per dinàmiques complexes, grans retards temporals i pendents negligibles, aspectes que en dificulten la gestió. Per tal d’assolir la ges tió òptima, s’ha de garantir un control eficient de les estructures hidràuliques tals com comportes, dics i rescloses. Amb aquesta finalitat, es deriva un modelat del sistema orientat a control basat en un model existent simplificat, obtingut a partir de les equacions de Saint-Venant. Aquesta nova representació redueix la complexitat del model original, proporciona flexibilitat i permet coordinar informació actual i retardada de manera sistemàtica. Malgrat això, la formulació resultant pertany a la classe de sistemes descriptors amb retard, per als quals les tècniques de control i d’estimació estàndards necessiten ser esteses. En canvi, el control predictiu basat en models i l’estimació d’estat amb horitzó lliscant es poden adaptar fàcilment a la formulació proposada. A més, són capaços de tractar amb restriccions físiques i operacionals de forma natural. Degut a les grans dimensions de les vies de navegació interior, una implementació centralitzada no resulta, tot sovint, ni possible ni desitjada. Per tal de pal·liar aquest problema, es consideren mètodes no centralitzats. D’aquesta manera, es descompon el sistema global en subsistemes i es distribueix la càrrega computacional del problema centralitzat entre els agents locals, de manera que cadascun d’ells s’encarrega de fer complir els objectius locals . En tant que les vies de navegació interior són sistemes fortament connectats, se segueix un plantejament distribuït, incloent un protocol de comunicació entre els agents locals. Malgrat la optimalitat dels resultats que les estratègies proposades puguin proporcionar, l’estimació d’estat només serà efectiva a condició que els sensors proveeixin informació fiable. Igualment, les accions de control únicament es podran aplicar correctament si els actuadors no estan afectats per fallades. En efecte, qualsevol error pot conduir a una gestió ineficaç del sistema. És per aquest motiu que la darrera part de la tes i tracta s obre el disseny d’estratègies de supervisió, que permetin detectar i aïllar fallades en vies de navegació interior. Tots els resultats de modelat, control i estimació d’es tat centralitzats i distribuïts, així com de diagnòstic de fallades, s’apliquen a un cas d’estudi realista, basat en les vies de navegació interior del nord de França, per tal de provar-ne la seva eficàcia.
La presente tesis versa sobre el problema de la gestión óptima de los recursos hídricos en vías de navegación interior desde la perspectiva de la teoría de control. En concreto, el objetivo principal consiste en garantizar la condición de navegabilidad del sistema, es decir, garantizar que los niveles de agua de los canales sean tales que las embarcaciones puedan navegar de forma segura. Dicho objetivo se consigue manteniendo los niveles dentro de un intervalo alrededor del punto de operación. Otros objetivos comunes consisten en minimizar los costes asociados a la operación de los equipos, así como a extender su vida útil. Hay que tener en cuenta que las vías de navegación interiores son sistemas a gran escala caracterizados por dinámicas complejas, grandes retardos temporales y pendientes prácticamente nulas, lo que dificulta su gestión. Para alcanzar la gestión óptima, se debe garantizar un control eficiente de las estructuras hidráulicas tales como compuertas, diques y esclusas, y para ello se deriva un modelado del sistema orientado a control, basado en un modelo simplificado ya existente, obtenido a partir de las ecuaciones de Saint-Venant. Esta nueva representación reduce la complejidad del modelo original, proporciona flexibilidad y permite coordinar información actual y retardada de forma sistemática. Sin embargo, la formulación resultante pertenece a la clase de sistemas descriptores con retardos, para los cuales las técnicas de control y de estimación de estado estándares necesitan ser extendidas. En cambio, el control predictivo basado en modelos y la estimación de estado con horizonte deslizante pueden ser fácilmente adaptadas para la formulación propuesta, además de permitir lidiar con restricciones físicas y operacionales de forma natural. Hay que tener en cuenta que, debido a las grandes dimensiones de las vías de navegación interior, una implementación centralizada no es, a menudo, ni posible ni deseada, y para paliar este problema se consideran los enfoques no centralizados. De este modo, se descompone el sistema global en subsistemas y se distribuye la carga computacional del problema centralizado entre los agentes locales, de manera que cada uno de ellos se encarga de cumplir los objetivos locales. Como las vías de navegación interior son sistemas fuertemente conectados, se sigue un enfoque distribuido, incluyendo un protocolo de comunicación entre los agentes. También se ha de considerar que la estimación de estado sólo será efectiva a condición de que los sensores provean información fiable. Asimismo, las acciones de control únicamente se podrán aplicar correctamente si los actuadores no están afectados por fallas. En efecto, cualquier avería puede conducir a una gestión ineficaz del sistema. Es por ello que la última parte de la tesis trata sobre el diseño de estrategias de supervisión que permitan detectar y aislar fallas en vías de navegación interior. Todos los resultados de modelado, control y estimación de estado centralizados y distribuidos, así como de diagnóstico de fallas, se aplican a un caso de estudio realista basado en las vías de navegación interior del norte de Francia para probar su eficacia.

In order to meet tight product quality specifications for chemical batch processes, it is vital to monitor and control product quality throughout the batch duration. However, the frequent lack of in situ sensors for continuous monitoring of batch product quality complicates the control problem and calls for novel control approaches. This thesis focuses on the study and application of multivariate statistical methods to control product quality in chemical batch processes. These multivariate statistical methods can be used to identify data-driven prediction models that can be integrated within a model predictive control (MPC) framework. The ideal MPC control strategy achieves end-product quality specifications by performing trajectory tracking during the batch operating time. However, due to the lack of in-situ sensors, measurements of product quality are usually obtained by laboratory assays and are, therefore, inherently intermittent. This thesis proposes a new approach to realise trajectory tracking control of batch product quality in those situations where only intermittent measurements are available. The scope of this methodology consists of: 1) the identification of a partial least squares (PLS) model that works as an estimator of product quality, 2) the transformation of the PLS model into a recursive formulation utilising a moving window technique, and 3) the incorporation of the recursive PLS model as a predictor into a standard MPC framework for tracking the desired trajectory of batch product quality. The structure of the recursive PLS model allows a straightforward incorporation of process constraints in the optimisation process. Additionally, a method to incorporate a nonlinear inner relation within the proposed PLS recursive model is introduced. This nonlinear inner relation is a combination of feedforward artificial neural networks (ANNs) and linear regression. Nonlinear models based on this method can predict product quality of highly nonlinear batch processes and can, therefore, be used within an MPC framework to control such processes. The use of linear regression in addition to ANNs within the PLS model reduces the risk of overfitting and also reduces the computational e↵ort of the optimisation carried out by the controller. The benefits of the proposed modelling and control methods are demonstrated using a number of simulated batch processes.

The capacity expansion of a district heating system is studied with the objective of evaluating the investment decision timing and type of capacity expansion. District energy is an energy generation system that provides energy, such as heat and electricity, generated at central locations and distributed to the surrounding area. The study develops an optimization framework to find the optimal investment schedule over a 30 year horizon with the options of investing in traditional heating sources (boilers) or a next-generation combined heat and power (CHP) plant that can provide heat and electricity. In district energy systems, the investment decision on the capacity and type of system is dependent on demand-side requirements, energy prices, and environmental costs. The main contribution of this work is to formulate the capacity planning over a time horizon asa dynamic optimal control problem. In this way, an initial system configuration can be modified by a 'controller' that optimally applies control actions that drive the system from an initial state to an optimal state. The optimal control is a model predictive control (MPC) formulation that not only provides the timing and size of the capacity investment, but also guidance on the mode of operation that meets optimal economic objectives with the given capacity.

In the process industries, cooling capacity is an important enabler for the facility to manufacture on specification product. The cooling water network is an important part of the over-all cooling system of the facility. In this research a cooling water circuit consisting of 3 cooling towers in parallel, 2 cooling water pumps in parallel, and 11 heat exchangers in parallel, is modelled. The model developed is based on first principles and captures the dynamic, non-linear, interactive nature of the plant. The modelled plant is further complicated by continuous, as well as discrete process variables, giving the model a hybrid nature. Energy consumption is included in the model as it is a very important parameter for plant operation. The model is fitted to real industry data by using a particle swarm optimisation approach. The model is suitable to be used for optimisation and control purposes. Cooling water networks are often not instrumented and actuated, nor controlled or optimised. Significant process benefits can be achieved by better process end-user temperature control, and direct monetary benefits can be obtained from electric power minimisation. A Hybrid Non-Linear Model Predictive Control strategy is developed for these control objectives, and simulated on the developed first principles dynamic model. Continuous and hybrid control cases are developed, and tested on process scenarios that reflect conditions seen in a real plant. Various alternative techniques are evaluated in order to solve the Hybrid Non-Linear Control problem. Gradient descent with momentum is chosen and configured to be used to solve the continuous control problem. For the discrete control problem a graph traversal algorithm is developed and joined to the continuous control algorithm to form a Hybrid Non-Linear Model Predictive controller. The potential monetary benefits that can be obtained by the plant owner through implementing the designed control strategy, are estimated. A powerful computation platform is designed for the plant model and controller simulations.
Thesis (PhD)--University of Pretoria, 2019.
Electrical, Electronic and Computer Engineering
PhD
Unrestricted

This research work provides the development of an Adaptive Flight Control System (AFCS) for autonomous hover of a Rotary-wing Unmanned Aerial Vehicle (RUAV). Due to the complex, nonlinear and time-varying dynamics of the RUAV, indirect adaptive control using the Model Predictive Control (MPC) is utilised. The performance of the MPC mainly depends on the model of the RUAV used for predicting the future behaviour. Due to the complexities associated with the RUAV dynamics, a neural network based black box identification technique is used for modelling the behaviour of the RUAV. Auto-regressive neural network architecture is developed for offline and online modelling purposes. A hybrid modelling technique that exploits the advantages of both the offline and the online models is proposed. In the hybrid modelling technique, the predictions from the offline trained model are corrected by using the error predictions from the online model at every sample time. To reduce the computational time for training the neural networks, a principal component analysis based algorithm that reduces the dimension of the input training data is also proposed. This approach is shown to reduce the computational time significantly. These identification techniques are validated in numerical simulations before flight testing in the Eagle and RMAX helicopter platforms. Using the successfully validated models of the RUAVs, Neural Network based Model Predictive Controller (NN-MPC) is developed taking into account the non-linearity of the RUAVs and constraints into consideration. The parameters of the MPC are chosen to satisfy the performance requirements imposed on the flight controller. The optimisation problem is solved numerically using nonlinear optimisation techniques. The performance of the controller is extensively validated using numerical simulation models before flight testing. The effects of actuator and sensor delays and noises along with the wind gusts are taken into account during these numerical simulations. In addition, the robustness of the controller is validated numerically for possible parameter variations. The numerical simulation results are compared with a base-line PID controller. Finally, the NN-MPCs are flight tested for height control and autonomous hover. For these, SISO as well as multiple SISO controllers are used. The flight tests are conducted in varying weather conditions to validate the utility of the control technique. The NN-MPC in conjunction with the proposed hybrid modelling technique is shown to handle additional disturbances successfully. Extensive flight test results provide justification for the use of the NN-MPC technique as a reliable technique for control of non-linear complex dynamic systems such as RUAVs.

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.

This thesis presents sorne contributions to the field of Health-Aware Control (HAC) of dynamic systems. In the first part of this thesis, a review of the concepts and methodologies related to reliability versus degradation and fault tolerant control versus health-aware control is presented. Firstly, in an attempt to unify concepts, an overview of HAC, degradation, and reliability modeling including some of the most relevant theoretical and applied contributions is given. Moreover, reliability modeling is formalized and exemplified using the structure function, Bayesian networks (BNs) and Dynamic Bayesian networks (DBNs) as modeling tools in reliability analysis. In addition, some Reliability lmportance Measures (RIMs) are presented. In particular, this thesis develops BNs models for overall system reliability analysis through the use of Bayesian inference techniques. Bayesian networks are powerful tools in system reliability assessment due to their flexibility in modeling the reliability structure of complex systems. For the HAC scheme implementation, this thesis presents and discusses the integration of actuators health information by means of RIMs and degradation in Model Predictive Control (MPC) and Linear Quadratic Regulator algorithms. In the proposed strategies, the cost function parameters are tuned using RIMs. The methodology is able to avoid the occurrence of catastrophic and incipient faults by monitoring the overall system reliability. The proposed HAC strategies are applied to a Drinking Water Network (DWN) and a multirotor UAV system. Moreover, a third approach, which uses MPC and restricts the degradation of the system components is applied to a twin rotor system. Finally, this thesis presents and discusses two reliability interpretations. These interpretations, namely instantaneous and expected, differ in the manner how reliability is evaluated and how its evolution along time is considered. This comparison is made within a HAC framework and studies the system reliability under both approaches.
Aquesta tesi presenta algunes contribucions al camp del control basat en la salut dels components "Health-Aware Control" (HAC) de sistemes dinàmics. A la primera part d'aquesta tesi, es presenta una revisió dels conceptes i metodologies relacionats amb la fiabilitat versus degradació, el control tolerant a fallades versus el HAC. En primer lloc, i per unificar els conceptes, s'introdueixen els conceptes de degradació i fiabilitat, models de fiabilitat i de HAC incloent algunes de les contribucions teòriques i aplicades més rellevants. La tesi, a més, el modelatge de la fiabilitat es formalitza i exemplifica utilitzant la funció d'estructura del sistema, xarxes bayesianes (BN) i xarxes bayesianes dinamiques (DBN) com a eines de modelat i anàlisi de la fiabilitat com també presenta algunes mesures d'importància de la fiabilitat (RIMs). En particular, aquesta tesi desenvolupa models de BNs per a l'anàlisi de la fiabilitat del sistema a través de l'ús de tècniques d'inferència bayesiana. Les xarxes bayesianes són eines poderoses en l'avaluació de la fiabilitat del sistema gràcies a la seva flexibilitat en el modelat de la fiabilitat de sistemes complexos. Per a la implementació de l?esquema de HAC, aquesta tesi presenta i discuteix la integració de la informació sobre la salut i degradació dels actuadors mitjançant les RIMs en algoritmes de control predictiu basat en models (MPC) i control lineal quadràtic (LQR). En les estratègies proposades, els paràmetres de la funció de cost s'ajusten utilitzant els RIMs. Aquestes tècniques de control fiable permetran millorar la disponibilitat i la seguretat dels sistemes evitant l'aparició de fallades a través de la incorporació d'aquesta informació de la salut dels components en l'algoritme de control. Les estratègies de HAC proposades s'apliquen a una xarxa d'aigua potable (DWN) i a un sistema UAV multirrotor. A més, un tercer enfocament fent servir la degradació dels actuadors com a restricció dins l'algoritme de control MPC s'aplica a un sistema aeri a dos graus de llibertat (TRMS). Finalment, aquesta tesi també presenta i discuteix dues interpretacions de la fiabilitat. Aquestes interpretacions, nomenades instantània i esperada, difereixen en la forma en què s'avalua la fiabilitat i com es considera la seva evolució al llarg del temps. Aquesta comparació es realitza en el marc del control HAC i estudia la fiabilitat del sistema en tots dos enfocaments.
Esta tesis presenta algunas contribuciones en el campo del control basado en la salud de los componentes “Health-Aware Control” (HAC) de sistemas dinámicos. En la primera parte de esta tesis, se presenta una revisión de los conceptos y metodologíasrelacionados con la fiabilidad versus degradación, el control tolerante a fallos versus el HAC. En primer lugar, y para unificar los conceptos, se introducen los conceptos de degradación y fiabilidad, modelos de fiabilidad y de HAC incluyendo algunas de las contribuciones teóricas y aplicadas más relevantes. La tesis, demás formaliza y ejemplifica el modelado de fiabilidad utilizando la función de estructura del sistema, redes bayesianas (BN) y redes bayesianas diná-micas (DBN) como herramientas de modelado y análisis de fiabilidad como también presenta algunas medidas de importancia de la fiabilidad (RIMs). En particular, esta tesis desarrolla modelos de BNs para el análisis de la fiabilidad del sistema a través del uso de técnicas de inferencia bayesiana. Las redes bayesianas son herramientas poderosas en la evaluación de la fiabilidad del sistema gracias a su flexibilidad en el modelado de la fiabilidad de sistemas complejos. Para la implementación del esquema de HAC, esta tesis presenta y discute la integración de la información sobre la salud y degradación de los actuadores mediante las RIMs en algoritmos de control predictivo basado en modelos (MPC) y del control cuadrático lineal (LQR). En las estrategias propuestas, los parámetros de la función de coste se ajustan utilizando las RIMs. Estas técnicas de control fiable permitirán mejorar la disponibilidad y la seguridad de los sistemas evitando la aparición de fallos a través de la incorporación de la información de la salud de los componentes en el algoritmo de control. Las estrategias de HAC propuestas se aplican a una red de agua potable (DWN) y a un sistema UAV multirotor. Además, un tercer enfoque que usa la degradación de los actuadores como restricción en el algoritmo de control MPC se aplica a un sistema aéreo con dos grados de libertad (TRMS). Finalmente, esta tesis también presenta y discute dos interpretaciones de la fiabilidad. Estas interpretaciones, llamadas instantánea y esperada, difieren en la forma en que se evalúa la fiabilidad y cómo se considera su evolución a lo largo del tiempo. Esta comparación se realiza en el marco del control HAC y estudia la fiabilidad del sistema en ambos enfoques.

Networked predictive control is a new research method for Networked Control Systems (NCS), which is able to handle network-induced problems such as time-delay, data dropouts, packets disorders, etc. while stabilizing the closed-loop system. This work is an extension and complement of networked predictive control methodology. There is always present model uncertainties or physical nonlinearity in the process of NCS. Therefore, it makes the study of the robust control of NCS and that of networked nonlinear control system (NNCS) considerably important. This work studied the following three problems: the robust control of networked predictive linear control systems, the control scheme for networked nonlinear control systems (NNCS) and the robust control of NNCS. The emphasis is on stability analysis and the design of robust control. This work adapted the two control schemes, namely, the time-driven and the event driven predictive controller for the implementation of NCS. It studied networked linear control systems and networked nonlinear control systems. Firstly, time-driven predictive controller is used to compensate for the networked-induced problems of a class of networked linear control systems while robustly stabilizing the closed-loop system. Secondly, event-driven predictive controller is applied to networked linear control system and NNCS and the work goes on to solve the robust control problem. The event-driven predictive controller brings great benefits to NCS implementation: it makes the synchronization of the clocks of the process and the controller unnecessary and it avoids measuring the exact values of the individual components of the network induced time-delay. This work developed the theory of stability analysis and robust synthesis of NCS and NNCS. The robust stability analysis and robust synthesis of a range of different system configurations have been thoroughly studied. A series of methods have been developed to handle the stability analysis and controller design for NCS and NNCS. The stability of the closed-loop of NCS has been studied by transforming it into that of a corresponding augmented system. It has been proved that if some equality conditions are satisfied then the closed-loop of NCS is stable for an upper-bounded random time delay and data dropouts. The equality conditions can be incorporated into a sub-optimal problem. Solving the sub-optimal problem gives the controller parameters and thus enables the synthesis of NCS. To simplify the calculation of solving the controller parameters, this thesis developed the relationship between networked nonlinear control system and a class of uncertain linear feedback control system. It proves that the controller parameters of some types of networked control system can be equivalently derived from the robust control of a class of uncertain linear feedback control system. The methods developed in this thesis for control design and robustness analysis have been validated by simulations or experiments.

The main interest of the present thesis is the constrained control of time-delay system, more specifically taking into consideration the discretization problem (due to, for example, a communication network) and the presence of constraints in the system's trajectories and control inputs. The effects of data-sampling and modeling problem are studied in detail, where an uncertainty is added into the system due to additional effect of the discretization and delay. The delay variation with respect to the sampling instants is characterized by a polytopic supra-approximation of the discretization/delay induced uncertainty. Some stabilizing techniques, based on Lyapunov's theory, are then derived for the unconstrained case. Lyapunov-Krasovskii candidates were also used to obtain LMI conditions for a state feedback, in the ''original" state-space of the system. For the constrained control purposes, the set invariance theory is used intensively, in order to obtain a region where the system is ''well-behaviored", despite the presence of constraints and (time-varying) delay. Due to the high complexity of the maximal delayed state admissible set obtained in the augmented state-space approach, in the present manuscript we proposed the concept of set invariance in the ''original" state-space of the system, called D-invariance. Finally, in the las part of the thesis, the MPC scheme is presented, in order to take into account the constraints and the optimality of the control solution.

Non-centralized control schemes for large-scale systems, including energy networks, are more flexible, scalable, and reliable than the centralized counterpart. These benefrts are obtained by having a set of local control!ers, each of which is responsible for a partition of the system, instead of one central entity that controls the whole system. Furthermore,in sorne cases, employing a non­ centralized control structure might be necessary due to the intractability problem of the centralized method.Thus, this thesis is devoted to the study of non-centralized optimization-based control approaches for large-scale energy systems. Mainly,this thesis focuses on the communication and cooperation processes of local controllers, which are integral parts of such schemes. Throughout this thesis,the model predictíve control framework is applied to solve the economic dispatch problem of large-scale energy systems. In a non-centralized architecture, local controllers must cooperatively solve the economic dispatch problem, which is formulated as a convex optimization problem with edge-based coupling constraints, at each time step.Therefore, first, the augmented Lagrangian approach is deployed to decompose the problem and to design two distributed optimization methods, which are iterative and require the local controllers to exchange information with each other at each iteration. lt is then shown that the sequence produced by these methods converges to an optima!solution when sorne cond tions, which include how the controllers must communicate and cooperate, are satisfied. However, in practice, the communication process might not always be perfect,i.e.,the required communication assumption does not hold. In the case of communication link failures, the distributed methods might not be able to compute a solution.Therefore,an information exchange protocol that is based on consensus is designed to overcome this problem. Furthermore, the proposed distributed methods are also further·extended such that they work over random communication networks and asynchronous updates, i.e.,when not all controllers always perform the updates . Under this setup, the convergence and the convergence rate of the algorithms are shown. Additionally, the implementation of these distributed methods to an MPC-based economic dispatch is also presented. The discussion includes the techniques that can be used to reduce the number of iterat ions and the performance of the methods in a numerical study. Considering that the aforementioned methods are comrnunication-intensive, an alternative non-centralized scheme, which provides a trade-off between comrnunication intensity and suboptirnality,is proposed.The scheme consists of repartitioning the network online with the aim of obtaining self-sufficient subsystems, forming coalitions for subsystems that are not self-sufficient,and decomposing the economic dispatch problem of the system into coalition-based subproblems. In this scheme, each subsystem only communicates to the others that belong to the sarne coalition;thus, reducing communication. Especially when all subsystems are self-sufficient, exchanging information is not needed. Finally,a cooperation problem during the implementation of the decisions is discussed. Specifically, sorne subsystems do not cornply with the computed decisions to gain better performance at the cost of deteriorating the performance of the other subsystems.A resilient scheme that can cope with this problem is formulated.lt consists of a stochastic method to robustify the decisions against such adversaria! behavior and an identification and mitigation method that is based on hypothesis testing using Bayesian inference.The proposed scheme, in general,can mitigate the effect of non-
Los esquemas de control no centralizados aplicados a sistemas a gran escala, entre los que se incluyen las redes energéticas, son más flexibles, escalables y fiables que sus equivalentes centralizados. Dichos beneficios pueden obtenerse empleando un conjunto de controladores locales, donde cada uno de ellos es responsable de una parte del sistema, en lugar de una entidad central que controle la totalidad del sistema.Asimismo,el uso de una estructura de control no centralizada podría ser, en algunos casos, necesario, dado el problema de intratabilidad del método centralizado. Por consiguiente, la presente tesis trata sobre el estudio de enfoques de control no centralizados basados en optimización para redes energéticas a gran escala. Principalmente, esta tesis se centra en los procesos de comunicación y cooperación llevados a cabo por los controladores locales , que constituyen partes esenciales de dichos esquemas . A lo largo de esta tesis, el control predictivo basado en modelos se usa para resolver el problema de expedir energia en redes energéticas a gran escala desde un punto de vista económico. En arquitecturas no centralizadas, los controladores locales deben resolver dicho problema de forma cooperativa, el cual se formula como un problema de optimización convexo con restricciones de acoplamiento en los enlaces entre nodos, que debe ser resuelto en cada instante de tiempo. Para ello, el método de Lagrangiano aumentado se utiliza inicialmente para descomponer el problema y diseñar dos métodos de optimización distribuidos , que son iterativos y requieren que los controladores locales intercambien información entre ellos en cada iteración . A continuación, se muestra que la secuencia generada por estos métodos converge a la solución óptima a condición de que se cumplan ciertas condiciones,incluyendo cómo los controladores deben comunicarse y cooperar. Sin embargo, en la práctica,la comunicación no siempre es perfecta, es decir,el supuesto de comunicación requerido no se cumple. En el caso de fallos en los enlaces de comunicación, los métodos distribuidos podrían no ser capaces de proporcionar una solución. Para paliar este problema, se diseña un protocolo de información basado en consenso.l'v1ás aún, los métodos de optimización distribuidos se extienden a fin de que sean capaces de trabajar en redes con comunicaciones aleatorias y actualizaciones asíncronas, es decir,redes en que no todos los controladores realicen las actualizaciones . En esta configuración se muestran la convergencia y el orden de convergencia de dichos algoritmos. Se muestra, además, la implementación de estos métodos en el control predictivo económico basado en modelos para redes energéticas. La discusión incluye las técnicas que pueden usarse para reducir el número de iteraciones, así como el desempeño de los métodos, a través de un estudio numérico. Teniendo en cuenta que los métodos anteriormente mencionados requieren una comunicación intensa,se propone otro esquema no centralizado que proporciona un compromiso entre intensidad de comunicación y suboptimalidad . Dicha estrategia consiste en volver a particionar en línea el sistema con el objetivo de obtener subsistemas autosuficientes,formando coaliciones de subsistemas que no lo sean por separado,y descomponiendo el problema económico de expedición de energía en subproblemas de tipo coalicional. En este esquema ,cada subsistema se comunica únicamente con aquellos otros subsistemas que pertenezcan a la misma coalición, reduciendo asi el tráfico de comunicación. En particular, cuando todos los subsistemas son autosuficientes, el intercambio de información ya no es necesario. Finalmente,se considera el problema de la cooperación durante la implementación de las decisiones Específicamente, algunos subsistemas no acatan las decisiones tomadas con el fin de lograr un desempeño propio superior a expensas de empeorar el desempeño de otros subsistemas. Es por esto que, con el fin de lidiar con este problema, se propone un esquema resiliente, el cual consiste en un método estocástico para hacer las decisiones más robustas frente a tal comportamiento adverso, y un método de identificación y mitigación basado en evaluación de hipótesis usando inferencia bayesiana. En general, el esquema propuesto logra mitigar el efecto de los subsistemas incumplidores sobre el resto, y en un caso concreto, también permite identificar los subsistemas adversos.
Els esquemes de control no centralitzats aplicats a sistemes a gran escala, entre els quals s’inclouen les xarxes energètiques, són més flexibles, escalables i fiables que els seus equivalents centralitzats. Aquests beneficis es poden obtenir fent servir un conjunt de controladors locals, en què cadascun d’ells és responsable d’una part del sistema, en lloc d’una entitat central que controli la totalitat del sistema. Així mateix, l’ús d’una estructura de control no centralitzada podria ser, en alguns casos, necessari, donat el problema d’intractabilitat del mètode centralitzat. Per tant, la present tesi tracta sobre l’estudi d’enfocaments de control no centralitzats basats en optimització per a xarxes energètiques a gran escala. Principalment, aquesta tesi se centra en els processos de comunicació i cooperació duts a terme pels controladors locals, que constitueixen parts essencials d’aquests esquemes. Al llarg d’aquesta tesi, el control predictiu basat en models s’utilitza per a resoldre el problema d’expedició d’energia en xarxes energètiques a gran escala des d’un punt de vista econòmic. En arquitectures no centralitzades, els controladors locals han de resoldre aquest problema de forma cooperativa, formulat com un problema d’optimització convex amb restriccions d’acoblament en els enllaços entre nodes i que ha de ser resolt a cada instant de temps. A tal efecte, el mètode de Lagrangià augmentat s’utilitza inicialment per a descomposar el problema i dissenyar dos mètodes d’optimització distribuïts, que són iteratius i requereixen que els controladors locals intercanviïn informació entre ells a cada iteració. A continuació, es mostra que la seqüència generada per aquests mètodes convergeix a la solució òptima si es compleixen certes condicions, incloent la manera en què els controladors s’han de comunicar i cooperar. No obstant això, a la pràctica, la comunicació no és sempre perfecta, és a dir, el supòsit de comunicació perfecta no es compleix. En el cas de fallades en els enllaços de comunicació, els mètodes distribuïts podrien no ser capaços de proporcionar una solució. Per a resoldre aquest problema, es dissenya un protocol d’informació basat en consens. A més, els mètodes d’optimització distribuïts s’amplien per tal que siguin capaços de treballar en xarxes amb comunicacions aleatòries i actualitzacions asíncrones, és a dir, xarxes en què no tots els controladors realitzin les actualitzacions. En aquestes configuracions es mostren la convergència i l’ordre de convergència d’aquests algoritmes. A més, es mostra també la implementació d’aquests mètodes en el control predictiu econòmic basat en models per a xarxes energètiques. La discussió inclou les tècniques que es poden emprar per a reduir el nombre d’iteracions, així com el rendiment dels mètodes, fent servir un estudi numèric. Tenint en compte que els mètodes anteriorment esmentats requereixen una comunicació intensa, es proposa un altre esquema no centralitzat que proporciona un compromís entre intensitat de comunicació i suboptimalitat. Aquesta estratègia consisteix en tornar a particionar el sistema en línia amb l’objectiu d’obtenir subsistemes autosuficients, formant coalicions de subsistemes que no ho siguin per separat, i descomposant el problema econòmic d’expedició d’energia en subproblemes de tipus coalicional. En aquest esquema, cada subsistema es comunica únicament amb aquells altre subsistemes que pertanyin a la mateixa coalició, reduint així el trànsit de comunicació. En particular, quan tots els sistemes són autosuficients, l’intercanvi d’informació deixa de ser necessari. Finalment, es considera el problema de la cooperació durant la implementació de les decisions. Específicament, alguns subsistemes no acaten les decisions preses amb la finalitat de millorar el propi rendiment a costa de disminuir el d’altres subsistemes. És per això que, a fi de solucionar aquest problema, es proposa un esquema resilient, el qual consisteix en un mètode estocàstic per fer les decisions més robustes davant d’aquest comportament advers, i un mètode d’identificació i mitigació basat en evaluar hipòtesis utilitzant inferència bayesiana. En general, l’esquema proposat aconsegueix mitigar l’efecte que els subsistemes no obedients exerceixen sobre la resta, i en un cas concert, també permet identificar els subsistemes adversos.
ABSTRAKSI (Indfonesian) Skema kendali yang tidak tersentralisasi untuk sistem berskala besar, seperti sistem aringan energi, lebih fleksibel, skalabel, dan reliabel dibandingkan dengan skema tersentralisasi. Keuntungan ini diperoleh dari terdapatnya satu set pengendali lokal, yang hanya bertanggung jawab terhadap satu partisi dari sistem tersebut, daripada jika hanya terdapat satu entitas yang mengendalikan seluruh sistem. Bahkan dalam beberapa sistem, penerapan struktur kendali yang tidak tersentralisasi menjadi keharusan karena adanya permasalahan intraktabilitas dari metode tersentralisasi. Oleh karena itu, disertasi ini bertujuan untuk melakukan studi pada metode kendali berdasarkan optimisasi dengan struktur yang tidak tersentralisasi untuk sistem energi berskala besar. Khususnya, disertasi ini memfokuskan pada proses komunikasi dan kooperasi pengendali‐pengendali lokal, yang merupakan bagian integral dalam skema yang dimaksud. Pada disertasi ini, sistem kontrol prediktif (model predictive control (MPC)) diterapkan untuk menyelesaikan optimisasi economic dispatch pada sistem energi berskala besar. Dalam arsitektur yang tidak tersentralisasi, pengendali‐pengendali lokal harus menyelesaikan permasalahan economic dispatch secara kooperatif. Permasalahan economic dispatch ini diformulasikan sebagai optimisasi yang konveks dan memiliki konstrain terkopling. Oleh karena itu, pendekatan Lagrange yang teraugmentasi diterapkan untuk mendekomposisi permasalahan optimisasi terkait. Pendekatan ini juga digunakan untuk merancang dua metode optimisasi terdistribusi, yang iteratif dan mengharuskan pengendali‐pengendali lokal bertukar informasi satu sama lain pada setiap iterasi. Sekuensi yang dihasilkan dari kedua metode tersebut akan terkonvergensi pada suatu solusi yang optimal apabila beberapa kondisi, yang meliputi bagaimana pengendali harus berkomunikasi dan berkooperasi, terpenuhi. Namun, pada praktiknya, proses komunikasi yang terjadi mungkin tidak selalu sempurna, dalam hal ini asumsi pada proses komunikasi yang dibutuhkan tidak terpenuhi. Pada kasus kegagalan jaringan komunikasi, metode terdistribusi yang dirancang mungkin tidak dapat menemukan solusinya. Oleh karena itu, suatu protokol untuk pertukaran informasi yang berdasarkan pada konsensus dirancang untuk mengatasi permasalahan ini. Selanjutnya, dua metode terdistribusi yang telah dirancang juga dikembangkan lebih jauh sehingga metode‐metode tersebut dapat bekerja pada jaringan komunikasi stokastik dengan proses yang asinkron, yaitu proses dimana tidak semua pengendali selalu melakukan pembaruan. Dalam hal ini, konvergensi dan laju konvergensi dari metode yang dirancang dipertunjukkan. Selain itu, implementasi dari metode terdistribusi pada sistem economic dispatch berbasis MPC juga dibahas. Diskusi pada bagian ini mencakup beberapa teknik yang dapat digunakan untuk mengurangi jumlah iterasi dan performa dari metode‐metode yang dirancang pada suatu studi numerik. Dengan pertimbangan bahwa metode‐metode yang disebut sebelumnya membutuhkan komunikasi yang intensif, maka sebuah skema alternatif, yang memberikan trade‐off antara intensitas komunikasi dan suboptimalitas, juga dirancang. Skema ini terdiri dari repartisi sistem online yang bertujuan untuk mendapatkan subsistemsubsistem yang swasembada, pembentukan koalisi untuk subsistem‐subsistem yang tidak swasembada, dan dekomposisi permasalahan economic dispatch menjadi subproblem berbasis koalisi. Dalam skema ini, tiap subsistem hanya perlu berkomunikasi dengan subsistem‐subsistem lain yang berada pada koalisi yang sama; sehingga mengurangi aliran komunikasi. Jika semua subsistem yang terbentuk swasembada, maka pertukaran informasi tidak dibutuhkan sama sekali. Pada akhirnya, disertasi ini juga membahas mengenai suatu permasalahan koperasi dalam masa implementasi keputusan (solusi). Pada permasalahan kooperasi ini, terdapat beberapa subsistem yang tidak menuruti keputusan (solusi), misalnya dengan tujuan untuk mendapatkan kinerja yang lebih baik dan di saat yang bersamaan memperburuk kinerja subsistem lainnya. Maka, sebuah skema resilien yang dapat mengatasi permasalahan ini dirumuskan. Skema tersebut terdiri dari sebuah metode stokastik untuk merobustifikasi keputusan terhadap perilaku adversari dan sebuah metode identifikasi dan mitigasi yang berdasarkan pada pengujian hipotesis dengan menggunakan inferensi Bayes. Skema yang diusulkan, secara umum, dapat memitigasi pengaruh subsistem yang tidak patuh pada subsistem reguler, dan pada kasus tertentu, juga dapat mengidentifikasi subsistem yang menjadi adversari.

The closed-loop control of processes over networks has in recent years become an increasingly popular research topic. This is a very viable solution for a wide variety of applications due to the rapid developments in communication network technologies and the widespread expansion of network devices and users. The convergence of communication networks technologies and advanced control methods do have a great potential to replace traditional control systems. The research programme presented in this thesis led to a development of networked predictive control algorithms over wired local area networks, general packet radio service wireless networks and wireless local area networks. Since the network is taken as a part of a control system, the network-induced time delay and data dropout are unavoidable. How to compensate for these issues is the main challenge in designing control methodologies for networked control systems. Five solutions were presented in this thesis to address these problems and were termed as recursive predictive control method I, inner loop predictive control method, outer loop predictive control method, modified generalised predictive control method and recursive predictive control method II. Irrespective of the different implementations of the networked control methods used, there is a common structure for each method which consists of a predictive control generator, a network delay compensator, a buffer and a plant output predictor. The predictive control generator and network delay compensator were used to compensate for the network delay and data dropout in the forward channel. The network delay and data dropout in the feedback channel was compensated for by using the plant output predictor, buffer and network delay compensator. The relationship between the sampling rate, packet size, network delay and data dropout were examined by using a round trip time delay method. Two network delay measurement methods were also presented and analysed in this thesis. The results of the real-time measurement of the network delay were used in an offline simulation. A networked servo system was built to test the system performance for an approximately linear, open-loop stable system and a networked inverted pendulum system was used to illustrate the system performance for an open-loop unstable system. The stability of each method was also considered. In order to simplify the software development, the Matlab/Simulink/Real-time workshop integrated development suit was used in the practical control system. The simulation block diagram in the Simulink environment was translated to the standard C language by using the real-time workshop. The ARMLINUX-GCC 3.4.4 was used to compile the generated C language file into the executable file running on an embedded board. In order to monitor the status of the control system and change the parameters of the controller, a network-based supervisory program was also developed using Microsoft Visual C++ 6.0.

This dissertation intends to provide theoretical and practical contributions on estimation, diagnosis and control of complex systems, especially in the mathematical form of descriptor systems. The research is motivated by real applications, such as water networks and power systems, which require a control system to provide a proper management able to take into account their specific features and operating limits in presence of uncertainties related to their operation and failures from component malfunctions. Such a control system is expected to provide an optimal operation to obtain efficient and reliable performance. State estimation is an essential tool, which can be used not only for fault diagnosis but also for the controller design. To achieve a satisfactory robust performance, set theory is chosen to build a general framework for descriptor systems subject to uncertainties. Under certain assumptions, these uncertainties are propagated and bounded by deterministic sets that can be explicitly characterized at each iteration step. Moreover, set-invariance characterizations for descriptor systems are also of interest to describe the steady performance, which can also be used for active mode detection. For the controller design for complex systems, new developments of economic model predictive control (EMPC) are studied taking into account the case of underlying periodic behaviors. The EMPC controller is designed to be recursively feasible even with sudden changes in the economic cost function and the closed-loop convergence is guaranteed. Besides, a robust technique is plugged into the EMPC controller design to maintain these closed-loop properties in presence of uncertainties. Engineering applications modeled as descriptor systems are presented to illustrate these control strategies. From the real applications, some additional difficulties are solved, such as using a two-layer control strategy to avoid binary variables in real-time optimizations and using nonlinear constraint relaxation to deal with nonlinear algebraic equations in the descriptor model. Furthermore, the fault-tolerant capability is also included in the controller design for descriptor systems by means of the designed virtual actuator and virtual sensor together with an observer-based delayed controller.
Esta tesis propone contribuciones de carácter teórico y aplicado para la estimación del estado, el diagnóstico y el control óptimo de sistemas dinámicos complejos en particular, para los sistemas descriptores, incluyendo la capacidad de tolerancia a fallos. La motivación de la tesis proviene de aplicaciones reales, como redes de agua y sistemas de energía, cuya naturaleza crítica requiere necesariamente un sistema de control para una gestión capaz de tener en cuenta sus características específicas y límites operativos en presencia de incertidumbres relacionadas con su funcionamiento, así como fallos de funcionamiento de los componentes. El objetivo es conseguir controladores que mejoren tanto la eficiencia como la fiabilidad de dichos sistemas. La estimación del estado es una herramienta esencial que puede usarse no solo para el diagnóstico de fallos sino también para el diseño del control. Con este fin, se ha decidido utilizar metodologías intervalares, o basadas en conjuntos, para construir un marco general para los sistemas de descriptores sujetos a incertidumbres desconocidas pero acotadas. Estas incertidumbres se propagan y delimitan mediante conjuntos que se pueden caracterizar explícitamente en cada instante. Por otra parte, también se proponen caracterizaciones basadas en conjuntos invariantes para sistemas de descriptores que permiten describir comportamientos estacionarios y resultan útiles para la detección de modos activos. Se estudian también nuevos desarrollos del control predictivo económico basado en modelos (EMPC) para tener en cuenta posibles comportamientos periódicos en la variación de parámetros o en las perturbaciones que afectan a estos sistemas. Además, se demuestra que el control EMPC propuesto garantiza la factibilidad recursiva, incluso frente a cambios repentinos en la función de coste económico y se garantiza la convergencia en lazo cerrado. Por otra parte, se utilizan técnicas de control robusto pata garantizar que las estrategias de control predictivo económico mantengan las prestaciones en lazo cerrado, incluso en presencia de incertidumbre. Los desarrollos de la tesis se ilustran con casos de estudio realistas. Para algunas de aplicaciones reales, se resuelven dificultades adicionales, como el uso de una estrategia de control de dos niveles para evitar incluir variables binarias en la optimización y el uso de la relajación de restricciones no lineales para tratar las ecuaciones algebraicas no lineales en el modelo descriptor en las redes de agua. Finalmente, se incluye también una contribución al diseño de estrategias de control con tolerancia a fallos para sistemas descriptores.

The aim of this master thesis was to develop a versatile and reliable experimentalplatform of mobile robots, solving tracking problems, for education and research.Evaluation of parallel bearings-only tracking and control algorithms on a multi-corearchitecture has been performed. The platform was implemented as a mobile wirelesssensor network using multiple mobile robots, each using a mounted camera for dataacquisition. Data processing was performed on the mobile robots and on a server,which also played the role of network communication hub. A major focus was toimplement this platform in a flexible manner to allow for education and futureresearch in the fields of signal processing, wireless sensor networks and automaticcontrol. The implemented platform was intended to act as a bridge between the idealworld of simulation and the non-ideal real world of full scale prototypes.The implemented algorithms did estimation of the positions of the robots, estimationof a non-cooperating target's position and regulating the positions of the robots. Thetracking algorithms implemented were the Gaussian particle filter, the globallydistributed particle filter and the locally distributed particle filter. The regulator triedto move the robots to give the highest possible sensor information under givenconstraints. The regulators implemented used model predictive control algorithms.Code for communicating with filters in external processes were implementedtogether with tools for data extraction and statistical analysis.Both implementation details and evaluation of different tracking algorithms arepresented. Some algorithms have been tested as examples of the platformscapabilities, among them scalability and accuracy of some particle filtering techniques.The filters performed with sufficient accuracy and showed a close to linear speedupusing up to 12 processor cores. Performance of parallel particle filtering withconstraints on network bandwidth was also studied, measuring breakpoints on filtercommunication to avoid weight starvation. Quality of the sensor readings, networklatency and hardware performance are discussed. Experiments showed that theplatform was a viable alternative for data acquisition in algorithm development and forbenchmarking to multi-core architecture. The platform was shown to be flexibleenough to be used a framework for future algorithm development and education inautomatic control.

A predictive outage compensator co-located with the actuator node in a networked control system can be used to counteract unpredictable losses of data in the feedback control loop. When a new control command is not received at the actuator node at an appropriate time instance, the predictive outage compensator suggests a replacement command based on the history of past control commands. It is shown that a simple tuning phase together with the monitoring of the control history can lead to a compensator that can improve the closed-loop control performance under communication outages considerably compared to traditional schemes. Worst case performance bounds are given that relate the quality of the tuning phase and the complexity of the compensator with the length of the communication outage period. Zero-order-hold (holding the past control command if the current is lost) and applying an a priori decided constant signal (using a redefined value on the control command if the current is lost) are special cases of the more general compensation scheme presented. The predictive outage compensator is illustrated through computer simulation with communication outages.

QC 20120112