Dissertations / Theses on the topic 'Algebraic control systems'

The mathematical formulation of various control synthesis problems (such as Decentralized Stabilization Problem (DSP), Total Finite Settling Time Stabilization for discrete time linear systems (TFSTS), Exact Model Matching Problem (EMMP), Decoupling and Noninteracting Control Problems) via the algebraic framework of Matrix Fractional Representation (MFR) - i.e. the representation of the transfer matrices of the system as matrix fractions over the ring of interest - results to the study of matrix equations over rings, such as : A . X + B . Y = C , (X. A + Y . B =C) (1) A· X = B , (y. A = B) (2) A·X·B = C (3) A·X + Y·B = C, X·A + B·Y = C, A·X·B + C·Y·D = E (4). The main objective of this dissertation is to further investigate conditions for existence and characterization of certain types of solutions of equation (1) ; develop a unifying algebraic approach for solvability and characterization of solutions of equations (1) - (4), based on structural properties of the given matrices, over the ring of interest. The standard matrix Diophantine equation (1) is associated with the TFSTS for discrete time linear systems and issues concerning the characterization of solutions according to the Extended McMillan Degree (EMD) (minimum EMD, or fixed EMD) of the stabilizing controllers they define, are studied. A link between the issues in question and topological properties of certain families of solutions of (1) is established . Equation (1) is also studied in association with the DSP and Diagonal DSP (DDSP) for continuous time linear systems. Conditions for characterizing block diagonal solutions of (1) (which define decentralized stabilizing controllers) are derived and a closed form description of the families of diagonal and two blocks diagonal decentralized stabilizing controllers is introduced. The set of matrix equations (1) - (4) is assumed over the field of fractions of the ring of interest , ℛ , (mainly a Euclidean Domain (ED) and thus a Principal Ideal Domain (PID)) and solvability as well as parametrization of solutions over ℛ is investigated under the unifying algebraic framework of extended non square matrix divisors, projectors and annihilators of the known matrices over ℛ . In practice the ring of interest is either the ring of polynomials ℝ [s] , or the rings of proper ℝ_pr(s) and especially proper and stable rational functions R_op(s). The importance of R_op(s) is highlighted early in the thesis and further computational issues arising from its structure as an ED are considered.

Dans la première partie de cette thèse, nous étudions les équations différentielles algébriques (en abrégé EDA) linéaires et les systèmes de contrôles linéaires associés (en abrégé SCEDA). Les problèmes traités et les résultats obtenus sont résumés comme suit : 1. Relations géométriques entre les EDA linéaires et les systèmes de contrôles génériques SCEDO. Nous introduisons une méthode, appelée explicitation, pour associer un SCEDO à n'importe quel EDA linéaire. L'explicitation d'une EDA est une classe des SCEDO, précisément un SCEDO défini, à un changement de coordonnées près, une transformation de bouclage près et une injection de sortie près. Puis nous comparons les « suites de Wong » d'une EDA avec les espaces invariants de son explicitation. Nous prouvons que la forme canonique de Kronecker FCK d'une EDA linéaire et la forme canonique de Morse FCM d'un SCEDO, ont une correspondance une à une et que leurs invariants sont liés. De plus, nous définissons l'équivalence interne de deux EDA et montrons sa particularité par rapport à l'équivalence externe en examinant les relations avec la régularité interne, i.e., l'existence et l'unicité de solutions. 2. Transformation d'un SCEDA linéaire vers sa forme canonique via la méthode d'explicitation avec des variables de driving. Nous étudions les relations entre la forme canonique par bouclage FCFB d'un SCEDA proposée dans la littérature et la forme canonique de Morse pour les SCEDO. Premièrement, dans le but de relier SCEDA avec les SCEDO, nous utilisons une méthode appelée explicitation (avec des variables de driving). Cette méthode attache à une classe de SCEDO avec deux types d'entrées (le contrôle original et le vecteur des variables de driving) à un SCEDA donné. D'autre part, pour un SCEDO linéaire classique (sans variable de driving) nous proposons une forme de Morse triangulaire FMT pour modifier la construction de la FCM. Basé sur la FMT nous proposons une forme étendue FMT et une forme étendue de FCM pour les SCEDO avec deux types d'entrées. Finalement, un algorithme est donné pour transformer un SCEDA dans sa FCFB. Cet algorithme est construit sur la FCM d'un SCEDO donné par la procédure d'explicitation. Un exemple numérique illustre la structure et l'efficacité de l'algorithme. Pour les EDA non linéaires et les SCEDA (quasi linéaires) nous étudions les problèmes suivants : 3. Explicitations, analyse externe et interne et formes normales des EDA non linéaires. Nous généralisons les deux procédures d'explicitation (avec ou sans variables de driving) dans le cas des EDA non linéaires. L'objectif de ces deux méthodes est d'associer un SCEDO non linéaire à une EDA non linéaire telle que nous puissions l'analyser à l'aide de la théorie des EDO non linéaires. Nous comparons les différences de l'équivalence interne et externe des EDA non linéaires en étudiant leurs relations avec l'existence et l'unicité d'une solution (régularité interne). Puis nous montrons que l'analyse interne des EDA non linéaire est liée à la dynamique nulle en théorie classique du contrôle non linéaire. De plus, nous montrons les relations des EDAS de forme purement semi-explicite avec les 2 procédures d'explicitations. Finalement, une généralisation de la forme de Weierstrass non linéaire FW basée sur la dynamique nulle d'un SCEDO non linéaire donné par la méthode d'explicitation est proposée
In the first part of this thesis, we study linear differential-algebraic equations (shortly, DAEs) and linear control systems given by DAEs (shortly, DAECSs). The discussed problems and obtained results are summarized as follows. 1. Geometric connections between linear DAEs and linear ODE control systems ODECSs. We propose a procedure, named explicitation, to associate a linear ODECS to any linear DAE. The explicitation of a DAE is a class of ODECSs, or more precisely, an ODECS defined up to a coordinates change, a feedback transformation and an output injection. Then we compare the Wong sequences of a DAE with invariant subspaces of its explicitation. We prove that the basic canonical forms, the Kronecker canonical form KCF of linear DAEs and the Morse canonical form MCF of ODECSs, have a perfect correspondence and their invariants (indices and subspaces) are related. Furthermore, we define the internal equivalence of two DAEs and show its difference with the external equivalence by discussing their relations with internal regularity, i.e., the existence and uniqueness of solutions. 2. Transform a linear DAECS into its feedback canonical form via the explicitation with driving variables. We study connections between the feedback canonical form FBCF of DAE control systems DAECSs proposed in the literature and the famous Morse canonical form MCF of ODECSs. In order to connect DAECSs with ODECSs, we use a procedure named explicitation (with driving variables). This procedure attaches a class of ODECSs with two kinds of inputs (the original control input and the vector of driving variables) to a given DAECS. On the other hand, for classical linear ODECSs (without driving variables), we propose a Morse triangular form MTF to modify the construction of the classical MCF. Based on the MTF, we propose an extended MTF and an extended MCF for ODECSs with two kinds of inputs. Finally, an algorithm is proposed to transform a given DAECS into its FBCF. This algorithm is based on the extended MCF of an ODECS given by the explication procedure. Finally, a numerical example is given to show the structure and efficiency of the proposed algorithm. For nonlinear DAEs and DAECSs (of quasi-linear form), we study the following problems: 3. Explicitations, external and internal analysis, and normal forms of nonlinear DAEs. We generalize the two explicitation procedures (with or without driving variable) proposed in the linear case for nonlinear DAEs of quasi-linear form. The purpose of these two explicitation procedures is to associate a nonlinear ODECS to any nonlinear DAE such that we can use the classical nonlinear ODE control theory to analyze nonlinear DAEs. We discuss differences of internal and external equivalence of nonlinear DAEs by showing their relations with the existence and uniqueness of solutions (internal regularity). Then we show that the internal analysis of nonlinear DAEs is closely related to the zero dynamics in the classical nonlinear control theory. Moreover, we show relations of DAEs of pure semi-explicit form with the two explicitation procedures. Furthermore, a nonlinear generalization of the Weierstrass form WE is proposed based on the zero dynamics of a nonlinear ODECS given by the explicitation procedure

The motion of fluids, such as air or water, is central to many engineering systems of significant economic and environmental importance. Examples range from air/fuel mixing in combustion engines to turbulence induced noise and fatigue on aircraft. Recent advances in novel sensor/actuator technologies have raised the intriguing prospect of actively sensing and manipulating the motion of the fluid within these systems, making them ripe for feedback control, provided a suitable control model exists. Unfortunately, the models for many of these systems are described by nonlinear, partial differential-algebraic equations for which few, if any, controller synthesis techniques exist. In stark contrast, the majority of established control theory assumes plant models of finite (and typically small) state dimension, expressed as a linear system of ordinary differential equations. Therefore, this thesis explores the problem of how to apply the mainstream tools of control theory to the class of systems described by partial differential-algebraic equations, that are either linear, or for which a linear approximation is valid. The problems of control system design for infinite-dimensional and algebraically constrained systems are treated separately in this thesis. With respect to the former, a new method is presented that enables the computation of a bound on the n-gap between a discretisation of a spatially distributed plant, and the plant itself, by exploiting the convergence rate of the v-gap metric between low-order models of successively finer spatial resolution. This bound informs the design, on loworder models, of H[infinity] loop-shaping controllers that are guaranteed to robustly stabilise the actual plant. An example is presented on a one-dimensional heat equation. Controller/estimator synthesis is then discussed for finite-dimensional systems containing algebraic, as well as differential equations. In the case of fluid flows, algebraic constraints typically arise from incompressibility and the application of boundary conditions. A numerical algorithm is presented, suitable for the semi-discrete linearised Navier-Stokes equations, that decouples the differential and algebraic parts of the system, enabling application of standard control theory without the need for velocity-vorticity type methods. This algorithm is demonstrated firstly on a simple electrical circuit, and secondly on the highly non-trivial problem of flow-field estimation in the transient growth region of a flat-plate boundary layer, using only wall shear measurements. These separate strands are woven together in the penultimate chapter, where a transient energy controller is designed for a channel-flow system, using wall mounted sensors and actuators.

Abstract In this thesis a set of new algorithms is introduced for Iterative Learning Control (ILC) and Repetitive Control (RC). Both areas of study are relatively new in control theory, and the common denominator for them is that they concentrate on controlling systems that include either reference signals or disturbances which are periodic. This provides opportunities for using past information or experience so that the control system learns the control action that results in good performance in terms of reference tracking or disturbance rejection. The first major contribution of the thesis is the algebraic analysis of ILC systems. This analysis shows that in the discrete-time case ILC algorithm design can be considered as designing a multivariable controller for a multivariable static plant and the reference signal that has to be tracked is a multivariable step function. Furthermore, the algebraic analysis reveals that time-varying algorithms should be used instead of time-invariant ones in order to guarantee monotonic convergence of the error in norm. However, from the algebraic analysis it is not clear how to select the free parameters of a given ILC algorithm. Hence in this thesis optimisation methods are used to automate this design phase. Special emphasis is placed on the so called Norm-Optimal Iterative Learning Control (NOILC) that was originally developed in (Amann:1996) as a new result it is shown that a convex modification of the existing predictive algorithm will result in a considerable improvement in convergence speed. Because the NOILC algorithm is computationally quite complex, a new set of Parameter-Optimal ILC algorithms are derived that converge under certain assumptions on the original plant. Three of these new algorithms will result in monotonic convergence to zero tracking error for an arbitrary discrete-time, linear, time-invariant plant. This a very strong property that has been earlier reported for only a small number of ILC algorithms. In the RC case it is shown that an existing RC algorithm that has been widely analysed and used in the research literature is in fact highly unrobust if the algorithm is implemented using sampled-data processing. Consequently, in this thesis a new optimality based discrete-time RC algorithm is derived, which converges to zero tracking error asymptotically for an arbitrary linear, time-invariant discrete-time plant under mild controllability and observability conditions.

I denna avhandling studeras optimal återkopplad styrning av olinjära deskriptorsystem. Ett deskriptorsystem är en matematisk beskrivning som kan innehålla både differentialekvationer och algebraiska ekvationer. En av anledningarna till intresset för denna klass av system är att objekt-orienterade modelleringsverktyg ger systembeskrivningar på denna form. Här kommer det att antas att det, åtminstone lokalt, är möjligt att eliminera de algebraiska ekvationerna och få ett system på tillståndsform. Teoretiskt är detta inte så inskränkande för genom att använda någon indexreduktionsmetod kan ganska generella deskriptor\-system skrivas om så att de uppfyller detta antagande.

För system på tillståndsform kan Hamilton-Jacobi-Bellman-ekvationen användas för att bestämma den optimala återkopplingen. Ett liknande resultat finns för deskriptor\-system där istället en Hamilton-Jacobi-Bellman-liknande ekvation ska lösas. Denna ekvation innehåller dock en extra term för att hantera de algebraiska ekvationerna. Eftersom antagandena i denna avhandling gör det möjligt att skriva om deskriptorsystemet som ett tillståndssystem, undersöks hur denna extra term måste väljas för att båda ekvationerna ska få samma lösning.

Ett problem med att beräkna den optimala återkopplingen med hjälp av Hamilton-Jacobi-Bellman-ekvationen är att det leder till att en olinjär partiell differentialekvation ska lösas. Generellt har denna ekvation ingen explicit lösning. Ett lättare problem är att beräkna en lokal optimal återkoppling. För analytiska system på tillståndsform löstes detta problem på 1960-talet och den optimala lösningen beskrivs av serieutvecklingar. I denna avhandling generaliseras detta resultat så att även deskriptor-system kan hanteras. Metoden illustreras med ett exempel som beskriver en faslåsande krets.

I många situationer vill man veta om ett område är möjligt att nå genom att styra på något sätt. För linjära tidsinvarianta system fås denna information från styrbarhetgramianen. För olinjära system används istället styrbarhetsfunktionen. Tre olika metoder för att beräkna styrbarhetsfunktionen har härletts i denna avhandling. De framtagna metoderna är också applicerade på några exempel för att visa beräkningsstegen.

Dessutom har observerbarhetsfunktionen studerats. Observerbarhetsfunktionen visar hur mycket utsignalenergi ett visst initial tillstånd svarar mot. Ett par olika metoder för att beräkna observerbarhetsfunktionen för deskriptorsystem tagits fram. För att beskriva en av metoderna, studeras ett litet exempel bestående av en elektrisk krets.

In this thesis, optimal feedback control for nonlinear descriptor systems is studied. A descriptor system is a mathematical description that can include both differential and algebraic equations. One of the reasons for the interest in this class of systems is that several modern object-oriented modeling tools yield system descriptions in this form. Here, it is assumed that it is possible to rewrite the descriptor system as a state-space system, at least locally. In theory, this assumption is not very restrictive because index reduction techniques can be used to rewrite rather general descriptor systems to satisfy this assumption.

The Hamilton-Jacobi-Bellman equation can be used to calculate the optimal feedback control for systems in state-space form. For descriptor systems, a similar result exists where a Hamilton-Jacobi-Bellman-like equation is solved. This equation includes an extra term in order to incorporate the algebraic equations. Since the assumptions made here make it possible to rewrite the descriptor system in state-space form, it is investigated how the extra term must be chosen in order to obtain the same solution from the different equations.

A problem when computing the optimal feedback law using the Hamilton-Jacobi-Bellman equation is that it involves solving a nonlinear partial differential equation. Often, this equation cannot be solved explicitly. An easier problem is to compute a locally optimal feedback law. This problem was solved in the 1960's for analytical systems in state-space form and the optimal solution is described using power series. In this thesis, this result is extended to also incorporate descriptor systems and it is applied to a phase-locked loop circuit.

In many situations, it is interesting to know if a certain region is reachable using some control signal. For linear time-invariant state-space systems, this information is given by the controllability gramian. For nonlinear state-space systems, the controllabilty function is used instead. Three methods for calculating the controllability function for descriptor systems are derived in this thesis. These methods are also applied to some examples in order to illustrate the computational steps.

Furthermore, the observability function is studied. This function reflects the amount of output energy a certain initial state corresponds to. Two methods for calculating the observability function for descriptor systems are derived. To describe one of the methods, a small example consisting of an electrical circuit is studied.

Report code: LiU-TEK-LIC-2006:8

In this thesis we present theoretical and numerical results for a feedback control problem defined by a thermal fluid. The problem is motivated by recent interest in designing and controlling energy efficient building systems. In particular, we show that it is possible to locally exponentially stabilize the nonlinear Boussinesq Equations by applying Neumann/Robin type boundary control on a bounded and connected domain. The feedback controller is obtained by solving a Linear Quadratic Regulator problem for the linearized Boussinesq equations. Applying classical results for semilinear equations where the linear term generates an analytic semigroup, we establish that this Riccati-based optimal boundary feedback control provides a local stabilizing controller for the full nonlinear Boussinesq equations. In addition, we present a finite element Galerkin approximation. Finally, we provide numerical results based on standard Taylor-Hood elements to illustrate the theory.
Ph. D.

[Truncated abstract. Please see the pdf version for the complete text. Also, formulae and special characters can only be approximated here. Please see the pdf version of this abstract for an accurate reproduction.] This thesis treats two aspects of intelligent control: The first part is about long-term optimization by approximating dynamic programming and in the second part a specific class of a fast neural network, related to support vector machines (SVMs), is considered. The first part relates to approximate dynamic programming, especially in the framework of adaptive critic designs (ACDs). Dynamic programming can be used to find an optimal decision or control policy over a long-term period. However, in practice it is difficult, and often impossible, to calculate a dynamic programming solution, due to the 'curse of dimensionality'. The adaptive critic design framework addresses this issue and tries to find a good solution by approximating the dynamic programming process for a stationary environment. In an adaptive critic design there are three modules, the plant or environment to be controlled, a critic to estimate the long-term cost and an action or controller module to produce the decision or control strategy. Even though there have been many publications on the subject over the past two decades, there are some points that have had less attention. While most of the publications address the training of the critic, one of the points that has not received systematic attention is training of the action module.¹ Normally, training starts with an arbitrary, hopefully stable, decision policy and its long-term cost is then estimated by the critic. Often the critic is a neural network that has to be trained, using a temporal difference and Bellman's principle of optimality. Once the critic network has converged, a policy improvement step is carried out by gradient descent to adjust the parameters of the controller network. Then the critic is retrained again to give the new long-term cost estimate. However, it would be preferable to focus more on extremal policies earlier in the training. Therefore, the Calculus of Variations is investigated to discard the idea of using the Euler equations to train the actor. However, an adaptive critic formulation for a continuous plant with a short-term cost as an integral cost density is made and the chain rule is applied to calculate the total derivative of the short-term cost with respect to the actor weights. This is different from the discrete systems, usually used in adaptive critics, which are used in conjunction with total ordered derivatives. This idea is then extended to second order derivatives such that Newton's method can be applied to speed up convergence. Based on this, an almost concurrent actor and critic training was proposed. The equations are developed for any non-linear system and short-term cost density function and these were tested on a linear quadratic regulator (LQR) setup. With this approach the solution to the actor and critic weights can be achieved in only a few actor-critic training cycles. Some other, more minor issues, in the adaptive critic framework are investigated, such as the influence of the discounting factor in the Bellman equation on total ordered derivatives, the target interpretation in backpropagation through time as moving and fixed targets, the relation between simultaneous recurrent networks and dynamic programming is stated and a reinterpretation of the recurrent generalized multilayer perceptron (GMLP) as a recurrent generalized finite impulse MLP (GFIR-MLP) is made. Another subject in this area that is investigated, is that of a hybrid dynamical system, characterized as a continuous plant and a set of basic feedback controllers, which are used to control the plant by finding a switching sequence to select one basic controller at a time. The special but important case is considered when the plant is linear but with some uncertainty in the state space and in the observation vector, and a quadratic cost function. This is a form of robust control, where a dynamic programming solution has to be calculated. ¹Werbos comments that most treatment of action nets or policies either assume enumerative maximization, which is good only for small problems, except for the games of Backgammon or Go [1], or, gradient-based training. The latter is prone to difficulties with local minima due to the non-convex nature of the cost-to-go function. With incremental methods, such as backpropagation through time, calculus of variations and model-predictive control, the dangers of non-convexity of the cost-to-go function with respect to the control is much less than the with respect to the critic parameters, when the sampling times are small. Therefore, getting the critic right has priority. But with larger sampling times, when the control represents a more complex plan, non-convexity becomes more serious.

Doctor of Philosophy
Department of Electrical and Computer Engineering
Don Gruenbacher
Caterina Scoglio
Cyber physical systems emerge when physical systems are integrated with communication networks. In particular, communication networks facilitate dissemination of data among components of physical systems to meet key requirements, such as efficiency and reliability, in achieving an objective. In this dissertation, we consider one of the most important cyber physical systems: the smart grid. The North American Electric Reliability Corporation (NERC) envisions a smart grid that aggressively explores advance communication network solutions to facilitate real-time monitoring and dynamic control of the bulk electric power system. At the distribution level, the smart grid integrates renewable generation and energy storage mechanisms to improve reliability of the grid. Furthermore, dynamic pricing and demand management provide customers an avenue to interact with the power system to determine electricity usage that satisfies their lifestyle. At the transmission level, efficient communication and a highly automated architecture provide visibility in the power system; hence, faults are mitigated faster than they can propagate. However, higher levels of reliability and efficiency rely on the supporting physical communication infrastructure and the network technologies employed. Conventionally, the topology of the communication network tends to be identical to that of the power network. In this dissertation, however, we employ a Demand Response (DR) application to illustrate that a topology that may be ideal for the power network may not necessarily be ideal for the communication network. To develop this illustration, we realize that communication network issues, such as congestion, are addressed by protocols, middle-ware, and software mechanisms. Additionally, a network whose physical topology is designed to avoid congestion realizes an even higher level of performance. For this reason, characterizing the communication infrastructure of smart grids provides mechanisms to improve performance while minimizing cost. Most recently, algebraic connectivity has been used in the ongoing research effort characterizing the robustness of networks to failures and attacks. Therefore, we first derive analytical methods for increasing algebraic connectivity and validate these methods numerically. Secondly, we investigate impact on the topology and traffic characteristics as algebraic connectivity is increased. Finally, we construct a DR application to demonstrate how concepts from graph theory can dramatically improve the performance of a communication network. With a hybrid simulation of both power and communication network, we illustrate that a topology which may be ideal for the power network may not necessarily be ideal for the communication network. To date, utility companies are embracing network technologies such as Multiprotocol Label Switching (MPLS) because of the available support for legacy devices, traffic engineering, and virtual private networks (VPNs) which are essential to the functioning of the smart grid. Furthermore, this particular network technology meets the requirement of non-routability as stipulated by NERC, but these benefits are costly for the infrastructure that supports the full MPLS specification. More importantly, with MPLS routing and other switching technologies, innovation is restricted to the features provided by the equipment. In particular, no practical method exists for utility consultants or researchers to test new ideas, such as alternatives to IP or MPLS, on a realistic scale in order to obtain the experience and confidence necessary for real-world deployments. As a result, novel ideas remain untested. On the contrary, OpenFlow, which has gained support from network providers such as Microsoft and Google and equipment vendors such as NEC and Cisco, provides the programmability and flexibility necessary to enable innovation in next-generation communication architectures for the smart grid. This level of flexibility allows OpenFlow to provide all features of MPLS and allows OpenFlow devices to co-exist with existing MPLS devices. Therefore, in this dissertation we explore a low-cost OpenFlow Software Defined Networking solution and compare its performance to that of MPLS. In summary, we develop methods for designing robust networks and evaluate software defined networking for communication and control in cyber physical systems where the smart grid is the system under consideration.

Industrial robot manipulators are general-purpose machines used for industrial automation in order to increase productivity, flexibility, and quality. Other reasons for using industrial robots are cost saving, and elimination of heavy and health-hazardous work. Robot motion control is a key competence for robot manufacturers, and the current development is focused on increasing the robot performance, reducing the robot cost, improving safety, and introducing new functionalities. Therefore, there is a need to continuously improve the models and control methods in order to fulfil all conflicting requirements, such as increased performance for a robot with lower weight, and thus lower mechanical stiffness and more complicated vibration modes. One reason for this development of the robot mechanical structure is of course cost-reduction, but other benefits are lower power consumption, improved dexterity, safety issues, and low environmental impact.

This thesis deals with three different aspects of modeling and control of flexible, i.e., elastic, manipulators. For an accurate description of a modern industrial manipulator, the traditional flexible joint model, described in literature, is not sufficient. An improved model where the elasticity is described by a number of localized multidimensional spring-damper pairs is therefore proposed. This model is called the extended flexible joint model. This work describes identification, feedforward control, and feedback control, using this model.

The proposed identification method is a frequency-domain non-linear gray-box method, which is evaluated by the identification of a modern six-axes robot manipulator. The identified model gives a good description of the global behavior of this robot.

The inverse dynamics control problem is discussed, and a solution methodology is proposed. This methodology is based on a differential algebraic equation (DAE) formulation of the problem. Feedforward control of a two-axes manipulator is then studied using this DAE approach.

Finally, a benchmark problem for robust feedback control of a single-axis extended flexible joint model is presented and some proposed solutions are analyzed.

In this digital age, cryptography is largely built in computer hardware or software as discrete structures. One of the most useful of these structures is finite fields. In this thesis, we explore a variety of applications of the theory and applications of arithmetic and computation in finite fields in both the areas of cryptography and cryptanalysis. First, multiplication algorithms in finite extensions of prime fields are explored. A new algebraic description of implementing the subquadratic Karatsuba algorithm and its variants for extension field multiplication are presented. The use of cy- clotomic fields and Gauss periods in constructing suitable extensions of virtually all sizes for efficient arithmetic are described. These multiplication techniques are then applied on some previously proposed public key cryptosystem based on exten- sion fields. These include the trace-based cryptosystems such as XTR, and torus- based cryptosystems such as CEILIDH. Improvements to the cost of arithmetic were achieved in some constructions due to the capability of thorough optimisation using the algebraic description. Then, for symmetric key systems, the focus is on algebraic analysis and attacks of stream ciphers. Different techniques of computing solutions to an arbitrary system of boolean equations were considered, and a method of analysing and simplifying the system using truth tables and graph theory have been investigated. Algebraic analyses were performed on stream ciphers based on linear feedback shift registers where clock control mechanisms are employed, a category of ciphers that have not been previously analysed before using this method. The results are successful algebraic attacks on various clock-controlled generators and cascade generators, and a full algebraic analyses for the eSTREAM cipher candidate Pomaranch. Some weaknesses in the filter functions used in Pomaranch have also been found. Finally, some non-traditional algebraic analysis of stream ciphers are presented. An algebraic analysis on the word-based RC4 family of stream ciphers is performed by constructing algebraic expressions for each of the operations involved, and it is concluded that each of these operations are significant in contributing to the overall security of the system. As far as we know, this is the first algebraic analysis on a stream cipher that is not based on linear feedback shift registers. The possibility of using binary extension fields and quotient rings for algebraic analysis of stream ciphers based on linear feedback shift registers are then investigated. Feasible algebraic attacks for generators with nonlinear filters are obtained and algebraic analyses for more complicated generators with multiple registers are presented. This new form of algebraic analysis may prove useful and thereby complement the traditional algebraic attacks. This thesis concludes with some future directions that can be taken and some open questions. Arithmetic and computation in finite fields will certainly be an important area for ongoing research as we are confronted with new developments in theory and exponentially growing computer power.

Conselho Nacional de Desenvolvimento Científico e Tecnológico
Optimal Control Problems (OCP), also known as Dynamic Optimization Problems, consist of an Objective Function to be maximized or minimized, associated with a set of differential and algebraic equations which include equality and inequality constraints in the state or control variables and characterize a system of Differential-Algebraic Equations (DAE). The differential-algebraic approach of numerical solution widely used in process simulation due the guarantee of attendance of the implicit algebraic constraints in the original formulation and the elimination of the necessary manipulations to transform the original problem into a purely differential system,was extended to OCP characterizing the called Differential-Algebraic Optimal Control Problem (DAOCP). A category of DAOCP of special interest includes inequality constraints, due the necessity of previous knowledge of the activations and deactivations sequence of these constraints along the trajectory and also of the instants where they occur, named Events. This DAOCPs with inequality constraints is equivalent to a class of hybrid dynamic optimization problems, where continuous and discrete behaviors are associated (FEEHERY, 1998). A particular type of hybrid OCP is that one where continuous state does not present jumps in the Events, called Switched OCP, for which Xu e Antsaklis (2004) considers a solution methodology based on the parameterization of Events with a previous specification of active subsystems sequence, resulting in the solution of a two-point boundary value differential-algebraic problem, formed by the state, co-state and stationarity equations, boundary and continuity conditions and its differentiations, called sensitivity equations. In this work, this indirect approach for Switched OCP was extended for DAOCP with inequality constraints, with the objective to estimate the Events, along the control, state and adjoint variables. The developed approach for Switched OCP described by Xu e Antsaklis (2004) was implemented in a specific code using Maple 9.5, called EVENTS, with the objective to symbolically generate the equations based on the parameterization of Events. This code was incorporated in a interface named OpCol, that collect characterization tools of DAE systems and generation of the optimality conditions extended Pontryagin s Principle for PCOAD of different types. The characterization tools are the INDEX of Murata (1996) that symbolically identifies the index, the resolubility and the consistency of initial conditions and the ACIG of Cunha e Murata (1999) that implements the Gear s algorithm for the index reduction and the index 1 equivalent system generation. The OTIMA (GOMES, 2000; LOBATO, 2004) generates the Euler-Lagrange equations for DAOCP. These tools had been implemented initially in different versions of Maple and all had been update to 9.5 version using the Maplets package that allows the data entry through interactive windows with the user, demanding a little knowledge of the Maple syntax. The OpCol interface was tested for four cases and for each tool a example data bank with typical problems of literature was created to assist the user in its use. Moreover, the direct method implemented in DIRCOL code was extended for multi-phases formulation with estimates of Events and the indirect method with Events Parameterization and differential-algebraic approach implemented in a Matlab code had been used for the numerical solution of three cases: a switched OCP and 2 DAOCP of batch reactors where the control variable is the feed rate of the component B - the first one has parallel reactions and selectivity constraints with 3 phases of index 1, 3 and 1 and the second a safety constraint with 2 phases of index 2 and 1 respectively and had been described by Srinivasan et al. (2003). The methodology used by this authors was applied to attained analytical expressions for the control variable in each phase necessary in indirect method, composing the called Switching Functions, from the optimality conditions based in the Pontryagin s Principle - specifically from the stationarity condition and the active constraint identification that will allow the control variable determination - and of the physical analysis of the problem in order to discard not appropriate activations/deactivations sequences. The results obtained by indirect and direct methods are compared for the 3 cited problems, showing the viability as much of the multiphase formulation using the DIRCOL and also the satisfactory performance of the indirect method with estimates of Events, beyond the utility of the tools of characterization of EADs, of attainment of optimality conditions and parameterization of Events available in Opcol interface.
Os Problemas de Controle Ótimo, também chamados Problemas de Otimização Dinâmica, são formados por uma Função Objetivo a ser maximizada ou minimizada, associada a conjuntos de equações algébricas e diferenciais que incluem restrições de igualdade e de desigualdade nas variáveis de estado e de controle que caracterizam um sistema de Equações Algébrico-Diferenciais (EADs). A extensão do ponto de vista algébricodiferencial de solução numérica aos PCOs, já amplamente utilizado na simulação de processos devido à garantia de atendimento às restrições algébricas originais e implícitas na formulação e à eliminação das manipulações necessárias para transformar o problema original num sistema de equações puramente diferenciais, caracteriza o chamado Problema de Controle Ótimo Algébrico-Diferencial (PCOAD). Uma categoria de PCOAD de especial interesse é a dos que incluem restrições de desigualdade, devido à necessidade de conhecimento prévio da seqüência de ativações e desativações destas restrições ao longo da trajetória e também dos instantes em que elas ocorrem, chamados Eventos. As ativações/desativações das restrições causam flutuações no índice diferencial e no número de graus de liberdade dinâmicos do PCOAD, exigindo técnicas especiais de redução deste índice até um e o emprego de métodos numéricos eficientes que garantam a convergência e estabilidade da solução. Estes PCOADs com restrições de desigualdade são equivalentes a uma classe de problemas de otimização dinâmica híbridos, que associam comportamentos contínuos e discretos (FEEHERY, 1998). Um tipo particular de PCO híbrido é aquele cujo estado contínuo não apresenta saltos nos Eventos, chamado PCO Chaveado, para o qual Xu e Antsaklis (2004) propõem uma metodologia de solução baseada na parametrização dos Eventos com a especificação prévia da seqüência de subsistemas ativos, resultando na solução de um problema de valor no contorno algébrico-diferencial em dois pontos, formado pelas equações de estado, co-estado e de estacionariedade, condições de contorno e de continuidade e suas diferenciações, chamadas equações de sensibilidade. Neste trabalho, esta abordagem indireta empregada para PCO Chaveados foi estendida para PCOAD com restrições de desigualdade, com o objetivo de estimar também os Eventos, além das variáveis de controle, de estado e adjuntas. A abordagem desenvolvida por Xu e Antsaklis (2004) para PCO Chaveados foi implementada num código específico utilizando o Maple 9.5, chamado EVENTS, com o objetivo de gerar simbolicamente as equações baseadas na parametrização dos Eventos. Este código foi incorporado a uma interface chamada OpCol, que reúne ferramentas de caracterização de sistemas de EAD e de geração das condições de otimalidade segundo o Princípio de Pontryagin estendidas para PCOAD de diferentes classes. As ferramentas de caracterização são o INDEX de Murata (1996) que identifica simbolicamente o índice, a resolubilidade e a consistência das condições iniciais e o ACIG de Cunha e Murata (1999) que implementa o algoritmo de Gear para a redução do índice e geração do sistema equivalente de índice 1. O OTIMA (GOMES, 2000; LOBATO, 2004) gera as equações de Euler-Lagrange para PCOAD. Estas ferramentas foram inicialmente implementadas em diferentes versões do Maple e todas foram atualizadas para a versão 9.5 utilizando o pacote Maplets que permite a entrada de dados através de janelas interativas com o usuário, exigindo dele pouco conhecimento da sintaxe Maple. A interface OpCol foi testada para quatro casos e para cada ferramenta foi criado um banco de exemplos com problemas típicos da literatura que auxiliam o usuário na sua utilização. Além disto, o método direto implementado no código DIRCOL estendido para formulações multifásicas com estimativa dos Eventos e o método indireto com Parametrização dos Eventos e abordagem algébrico-diferencial implementado num código MATLAB foram utilizados na solução numérica de três estudos de casos: um PCO chaveado e 2 PCOAD de reatores batelada onde a variável de controle é a taxa de alimentação do componente B: o primeiro tem reações paralelas e restrições de seletividade com 3 fases de índices 1, 3 e 1 e o segundo restrições de segurança com 2 fases de índices 2 e 1 e respectivamente e foram descritos por Srinivasan et al. (2003). A mesma metodologia utilizada por estes autores foi empregada na obtenção de expressões analíticas para a variável de controle em cada fase necessárias no método indireto, compondo as chamadas Funções Identificadoras de Fase (FIF), a partir das condições de otimalidade baseadas no Princípio de Pontryagin - especificamente a partir da condição de estacionariedade e da identificação da restrição ativa que permitirá a determinação da variável de controle - e da análise física do problema de modo a descartar seqüências de ativação/desativação não apropriadas. Os resultados obtidos pelo método indireto e pelo método direto são comparados entre si para os 3 problemas citados, mostrando a viabilidade tanto da formulação multifásica empregando o DIRCOL quanto o desempenho satisfatório do método indireto com estimativa de Eventos, além da utilidade das ferramentas de caracterização de EADs, de obtenção das condições de otimalidade e de parametrização dos eventos disponibilizadas na interface Opcol.
Mestre em Engenharia Química

This research effort develops a comprehensive computational framework to support the parametric optimal design of uncertain dynamical systems. Uncertainty comes from various sources, such as: system parameters, initial conditions, sensor and actuator noise, and external forcing. Treatment of uncertainty in design is of paramount practical importance because all real-life systems are affected by it; not accounting for uncertainty may result in poor robustness, sub-optimal performance and higher manufacturing costs. Contemporary methods for the quantification of uncertainty in dynamical systems are computationally intensive which, so far, have made a robust design optimization methodology prohibitive. Some existing algorithms address uncertainty in sensors and actuators during an optimal design; however, a comprehensive design framework that can treat all kinds of uncertainty with diverse distribution characteristics in a unified way is currently unavailable. The computational framework uses Generalized Polynomial Chaos methodology to quantify the effects of various sources of uncertainty found in dynamical systems; a Least-Squares Collocation Method is used to solve the corresponding uncertain differential equations. This technique is significantly faster computationally than traditional sampling methods and makes the construction of a parametric optimal design framework for uncertain systems feasible. The novel framework allows to directly treat uncertainty in the parametric optimal design process. Specifically, the following design problems are addressed: motion planning of fully-actuated and under-actuated systems; multi-objective robust design optimization; and optimal uncertainty apportionment concurrently with robust design optimization. The framework advances the state-of-the-art and enables engineers to produce more robust and optimally performing designs at an optimal manufacturing cost.
Ph. D.

Cette thèse porte sur la commande H∞ par loop-shaping pour les systèmes linéaires à temps invariant d'ordre faible avec ou sans retard et dépendant de paramètres inconnus. L'objectif est d'obtenir des correcteurs H∞ paramétriques, c'est-à-dire dépendant explicitement des paramètres inconnus, pour application à des viseurs gyrostabilisés.L'existence de ces paramètres inconnus ne permet plus l'utilisation des techniques numériques classiques pour la résolution du problème H∞ par loop-shaping. Nous avons alors développé une nouvelle méthodologie permettant de traiter les systèmes linéaires de dimension finie grâce à l'utilissation de techniques modernes de calcul formel dédiées à la résolution des systèmes polynomiaux (bases de Gröbner, variétés discriminantes, etc.).Une telle approche présente de multiples avantages: étude de sensibilités du critère H∞ par rapport aux paramètres, identification de valeurs de paramètres singulières ou remarquables, conception de correcteurs explicites optimaux/robustes, certification numérique des calculs, etc. De plus, nous montrons que cette approche peut s'étendre à une classe de systèmes à retard.Plus généralement, cette thèse s'appuie sur une étude symbolique des équations de Riccati algébriques. Les méthodologies génériques développées ici peuvent s'étendre à de nombreux problèmes de l'automatique, notamment la commande LQG, le filtrage de Kalman ou invariant
This PhD thesis deals with the H∞ loop-shaping design for low order linear time invariant systems depending on unknown parameters. The objective of the PhD thesis is to obtain parametric H∞ controllers, i.e. controllers which depend explicitly on the unknown model parameters, and to apply them to the stabilization of gyrostabilized sights.Due to the unknown parameters, no numerical algorithm can solve the robust control problem. Using modern symbolic techniques dedicated to the solving of polynomial systems (Gröbner bases, discriminant varieties, etc.), we develop a new methodology to solve this problem for finite-dimensional linear systems.This approach shows several advantages : we can study the sensibilities of the H∞ criterion to the parameter variations, identify singular or remarquable values of the parameters, compute controllers which depend explicitly on the parameters, certify the numerical computations, etc. Furthermore, we show that this approach can be extended to a class of linear time-delay systems.More generally, this PhD thesis develops an algebraic approach for the study of algebraic Riccati equations. Thus, the methodology obtained can be extended to many different problems such as LQG control and Kalman or invariant filtering

L’objectif principal de cette thèse est d’étudier le problème du contrôle de suivi distribué pour les systèmes de formation de multi-robots à contrainte non holonomique. Ce contrôle vise à entrainer une équipe de robots mobile de type monocycle pour former une configuration de formation désirée avec son centroïde se déplaçant avec une autre trajectoire de référence dynamique et pouvant être spécifié par le leader virtuel ou humain. Le problème du contrôle de suivi a été résolu au cours de cette thèse en développant divers contrôleurs distribués pratiques avec la considération d’un taux de convergence plus rapide, une précision de contrôle plus élevée, une robustesse plus forte, une estimation du temps de convergence explicite et indépendante et moins de coût de communication et de consommation d’énergie. Dans la première partie de la thèse nous étudions d’abord au niveau du chapitre 2 la stabilité à temps fini pour les systèmes de formation de multi-robots. Une nouvelle classe de contrôleur à temps fini est proposée dans le chapitre 3, également appelé contrôleur à temps fixe. Nous étudions les systèmes dynamiques de suivi de formation de multi-robots non holonomiques dans le chapitre 4. Dans la deuxième partie, nous étudions d'abord le mécanisme de communication et de contrôle déclenché par l'événement sur les systèmes de suivi de la formation de multi-robots non-holonomes au chapitre 5. De plus, afin de développer un schéma d'implémentation numérique, nous proposons une autre classe de contrôleurs périodiques déclenchés par un événement basé sur un observateur à temps fixe dans le chapitre 6
The main aim of this thesis is to study the distributed tracking control problem for the multi-robot formation systems with nonholonomic constraint, of which the control objective it to drive a team of unicycle-type mobile robots to form one desired formation configuration with its centroid moving along with another dynamic reference trajectory, which can be specified by the virtual leader or human. We consider several problems in this point, ranging from finite-time stability andfixed-time stability, event-triggered communication and control mechanism, kinematics and dynamics, continuous-time systems and hybrid systems. The tracking control problem has been solved in this thesis via developing diverse practical distributed controller with the consideration of faster convergence rate, higher control accuracy, stronger robustness, explicit and independent convergence time estimate, less communication cost and energy consumption.In the first part of the thesis, we first study the finite-time stability for the multi-robot formation systems in Chapter 2. To improve the pior results, a novel class of finite-time controller is further proposed in Chapter 3, which is also called fixed-time controller. The dynamics of nonholonomic multi-robot formation systems is considered in Chapter 4. In the second part, we first investigate the event-triggered communication and control mechanism on the nonholonomic multi-robot formation tracking systems in Chapter 5. Moreover, in order to develop a digital implement scheme, we propose another class of periodic event-triggered controller based on fixed-time observer in Chapter 6

Enforcing security policies to distributed systems is difficult, in particular, when a system contains untrusted components. We designed AspectKE*, a distributed AOP language based on a tuple space, to tackle this issue. In AspectKE*, aspects can enforce access control policies that depend on future behavior of running processes. One of the key language features is the predicates and functions that extract results of static program analysis, which are useful for defining security aspects that have to know about future behavior of a program. AspectKE* also provides a novel variable binding mechanism for pointcuts, so that pointcuts can uniformly specify join points based on both static and dynamic information about the program. Our implementation strategy performs fundamental static analysis at load-time, so as to retain runtime overheads minimal. We implemented a compiler for AspectKE*, and demonstrate usefulness of AspectKE* through a security aspect for a distributed chat system.

Dans cette thèse, on présente la théorie algébrique IH par le biais de générateurs et d’équations.Le modèle libre de IH est la catégorie des sous-espaces linéaires sur un corps k. Les termes de IH sont des diagrammes de cordes, qui, selon le choix de k, peuvent exprimer différents types de réseaux et de formalismes graphiques, que l’on retrouve dans des domaines scientifiques divers, tels que les circuits quantiques, les circuits électriques et les réseaux de Petri. Les équations de IH sont obtenues via des lois distributives entre algèbres de Hopf – d’où le nom “Interacting Hopf algebras” (algèbres de Hopf interagissantes). La caractérisation via les sous-espaces permet de voir IH comme une syntaxe fondée sur les diagrammes de cordes pour l’algèbre linéaire: les applications linéaires, les espaces et leurs transformations ont chacun leur représentation fidèle dans le langage graphique. Cela aboutit à un point de vue alternatif, souvent fructueux, sur le domaine.On illustre cela en particulier en utilisant IH pour axiomatiser la sémantique formelle de circuits de calculs de signaux, pour lesquels on s’intéresse aux questions de la complète adéquation et de la réalisabilité. Notre analyse suggère un certain nombre d’enseignements au sujet du rôle de la causalité dans la sémantique des systèmes de calcul
We present by generators and equations the algebraic theory IH whose free model is the category oflinear subspaces over a field k. Terms of IH are string diagrams which, for different choices of k, expressdifferent kinds of networks and graphical formalisms used by scientists in various fields, such as quantumcircuits, electrical circuits and Petri nets. The equations of IH arise by distributive laws between Hopfalgebras - from which the name interacting Hopf algebras. The characterisation in terms of subspacesallows to think of IH as a string diagrammatic syntax for linear algebra: linear maps, spaces and theirtransformations are all faithfully represented in the graphical language, resulting in an alternative, ofteninsightful perspective on the subject matter. As main application, we use IH to axiomatise a formalsemantics of signal processing circuits, for which we study full abstraction and realisability. Our analysissuggests a reflection about the role of causality in the semantics of computing devices

Esta proposta de dissertação de mestrado trata da aplicação de Estimadores Algébricos em sistemas de controle como alternativa ao uso de observadores. Devido à dificuldade de obtenção de resultados teóricos, dificuldade essa oriunda da natureza complexa dos estimadores algébricos, o trabalho é desenvolvido através do estudo de casos. Considera-se que a topologia de controle é a união de uma técnica tradicional de controle (por exemplo, uma realimentação de estado ou o método do torque calculado) com a estimação algébrica. Os resultados obtidos defendem a idéia de que os Estimadores Algébricos, quando usados como estimadores de estado, permitem obter um desempenho e uma robustez que se aproxima muito do desempenho e a robustez da mesma lei de controle no caso em que o estado é perfeitamente conhecido.
The control topology that is considered in this work is the union of a traditional control technique (e.g. a state feedback or the computed torque method) with the Algebraic Estimator. The obtained results reinforce the common sense about this class of estimators, that the use of Algebraic Estimators may produce the performance, robustness and noise immunity that mimics the case where perfect information of the state is available.

Orientador: Geraldo Nunes Silva
Banca: Antonio Carlos Gardel Leitão
Banca: Fernando Manuel Ferreira Lobo Pereira
Resumo: Neste trabalho é feito um estudo aprofundado da estabilidade de sistemas alternantes, principalmente via teoria de Lie. Inicialmente são apresentados os principais conceitos básicos da álgebra de Lie, necessários para o estudo dos critérios de estabilidade dos sistemas alternantes. Depois são discutidos critérios de estabilidade para sistemas alternantes. É feita a exposição da demonstração de que para todo sistema linear da forma ? x = Apx p = 1, 2, ...,N, com as matrizes Ap assintóticamente estáveis e comutativas duas a duas, existe uma função de Lyapunov quadrática comum. Uma condição suficiente para estabilidade assintótica de um sistema linear alternante é apresentada em termos da álgebra de Lie gerada por uma família infinita de matrizes. A saber, se esta álgebra de Lie é solúvel, então o sistema alternante é estável para uma mudança arbitrária de sinal. Em seguida são estudadas condições mais fracas. Supondo que a álgebra de Lie não é solúvel, mas é decomponível na soma de um ideal solúvel e uma subálgebra com grupo de Lie compacto, então o sistema alternante é globalmente exponencialmente uniformemente estável. Entretanto, se o grupo de Lie não for compacto, verifica-se que é possível gerar uma família finita de matrizes estáveis tais que o correspondente sistema linear alternante não é estável. Finalmente, os resultados correspondentes de estabilidade local para sistemas alternantes não lineares são apresentados.
Abstract: In this work it is undertaken a deep study of stability for switched systems, mainly via Lie algebraic Theory. At first, the basic concepts and results from Lie algebra necessary for the study of stability of switched systems are presented. Criteria for stability are discussed. It is also done an exposition of the proof that all linear systems ? x = Apx, p = 1, 2, ...,N, with stable and pairwisely commutative matrices Ap, have common quadratic Lyapounov functions. A sufficient condition for asymptotic stability of switched linear systems is presented in term of the Lie algebra generated by a family infinite matrices. That is, if this Lie algebra is solvable, then the switched systems are stable for an arbitrary change of sinal. Next weaker conditions are studied. If the Lie algebra is decomposable into two subalgebras in which one is a solvable ideal and the other has a compact Lie group, then the switched systems are globally exponentially uniformly stable. However, if the Lie group is not compact, it is also possible to generate a finite family of stable matrices such that the corresponding switched linear systems are not stable. Finally, corresponding local stability results are presented for nonlinear systems.
Mestre

We investigate the solution of large-scale generalized algebraic Bernoulli equations as those arising in control and systems theory in the context of stabilization of linear dynamical systems, coprime factorization of rational matrix-valued functions, and model reduction. The algorithms we propose, based on a generalization of the Newton iteration for the matrix sign function, are easy to parallelize, yielding an efficient numerical tool to solve large-scale problems. Both the accuracy and the parallel performance of our implementations on a cluster of Intel Xeon processors are reported.

O presente trabalho aborda o problema de estabilização local de sistemas não lineares quadráticos contínuos no tempo (possivelmente instáveis em malha aberta) e sujeitos a saturação de atuadores. Além disso o trabalho apresenta um estudo de técnicas de síntese de compensadores de anti-windup para sistemas quadráticos sujeitos à saturação de atuadores. A abordagem do estudo é comparativa em relação a duas formas de representação dos sistemas quadráticos. A primeira forma de abordagem é a Representação Algébrico-Diferencial — DAR (do inglês, Differential Algebraic Representation), aplicável a toda a classe de sistemas racionais. A segunda forma, por sua vez, consiste em uma decomposição quadrática, particular para sistemas quadráticos. Em ambos os casos, utiliza-se a não linearidade de zona morta e uma condição generalizada de setor para tratar da saturação. Para ambas representações, condições baseadas em Desigualdades Matriciais Lineares — LMIs (do inglês, Linear Matrix Inequalities) dependentes dos estados são obtidas para fornecer uma lei de controle linear, com o objetivo de estabilizar o sistema em malha fechada enquanto fornece uma região maximizada de estabilidade garantida associada a uma função de Lyapunov. A partir da mesma metodologia, são propostas técnicas de síntese de compensadores de anti-windup estáticos e dinâmicos. Exemplos numéricos são apresentados para verificar a eficácia dos métodos propostos.
This work addresses the problem of local stabilization of continuous-time quadratic systems (possibly open-loop unstable) and subject to actuator saturation. Furthermore, the work addresses a study of techniques for synthesis of anti-windup compensators for quadratic systems subject to actuator saturation. The study approach is comparative in the sense of considering two representations of quadratic systems. The first one is the Differential Algebraic Representation — DAR, suitable for the entire class of rational systems. The second representation consists in a quadratic decomposition, particular for quadratic systems. In both cases, it is used the deadzone nonlinearity and the generalized sector condition in order to deal with the saturation. For both representations, state-dependent Linear Matrix Inequalities — LMIs conditions are obtained to provide a control law with the aim of stabilize the closed-loop system while providing a region of guaranteed stability, associated to a Lyapunov function. Based on the same methodology, techniques are proposed for the synthesis of static and dynamic anti-windup compensators. Numerical examples are presented to verify the effectiveness of proposed methods.

Made available in DSpace on 2014-06-11T19:27:08Z (GMT). No. of bitstreams: 0 Previous issue date: 2005-09-05Bitstream added on 2014-06-13T20:16:05Z : No. of bitstreams: 1 nascimento_rp_me_sjrp.pdf: 368298 bytes, checksum: d1ffd79129c70e6a0b4236136ff5e58e (MD5)
Neste trabalho é feito um estudo aprofundado da estabilidade de sistemas alternantes, principalmente via teoria de Lie. Inicialmente são apresentados os principais conceitos básicos da álgebra de Lie, necessários para o estudo dos critérios de estabilidade dos sistemas alternantes. Depois são discutidos critérios de estabilidade para sistemas alternantes. É feita a exposição da demonstração de que para todo sistema linear da forma ? x = Apx p = 1, 2,...,N, com as matrizes Ap assintóticamente estáveis e comutativas duas a duas, existe uma função de Lyapunov quadrática comum. Uma condição suficiente para estabilidade assintótica de um sistema linear alternante é apresentada em termos da álgebra de Lie gerada por uma família infinita de matrizes. A saber, se esta álgebra de Lie é solúvel, então o sistema alternante é estável para uma mudança arbitrária de sinal. Em seguida são estudadas condições mais fracas. Supondo que a álgebra de Lie não é solúvel, mas é decomponível na soma de um ideal solúvel e uma subálgebra com grupo de Lie compacto, então o sistema alternante é globalmente exponencialmente uniformemente estável. Entretanto, se o grupo de Lie não for compacto, verifica-se que é possível gerar uma família finita de matrizes estáveis tais que o correspondente sistema linear alternante não é estável. Finalmente, os resultados correspondentes de estabilidade local para sistemas alternantes não lineares são apresentados.
In this work it is undertaken a deep study of stability for switched systems, mainly via Lie algebraic Theory. At first, the basic concepts and results from Lie algebra necessary for the study of stability of switched systems are presented. Criteria for stability are discussed. It is also done an exposition of the proof that all linear systems ? x = Apx, p = 1, 2, ...,N, with stable and pairwisely commutative matrices Ap, have common quadratic Lyapounov functions. A sufficient condition for asymptotic stability of switched linear systems is presented in term of the Lie algebra generated by a family infinite matrices. That is, if this Lie algebra is solvable, then the switched systems are stable for an arbitrary change of sinal. Next weaker conditions are studied. If the Lie algebra is decomposable into two subalgebras in which one is a solvable ideal and the other has a compact Lie group, then the switched systems are globally exponentially uniformly stable. However, if the Lie group is not compact, it is also possible to generate a finite family of stable matrices such that the corresponding switched linear systems are not stable. Finally, corresponding local stability results are presented for nonlinear systems.

L’objectif principal de cette thèse est l’étude de propriétés dynamiques et de méthodes de conception et synthèse des algorithmes de contrôle-commande des systèmes dans le cas où les fonctions de mesures, actionnements et contrôles sont distribuées sur des organes de calcul pouvant être partagés avec d’autres applications et connectés sur un réseau de communication numérique. En conséquence, les boucles de contrôle sont en compétition avec d’autres applications pour accéder aux ressources de calcul et de communication de capacité limitée et gérées par des politiques spécifiques. Ceci provoque l’apparition de délais et de perte d’informations transmises entre les différents nœuds qui peuvent dégrader les performances des systèmes et conduire à leur instabilité.Dans une première partie de la thèse, nous avons étudié l’analyse des performances de certains systèmes ainsi que la conception de contrôleurs robustes en fonction de la qualité de service fournie par le réseau. Cette étude a permis de spécifier les règles de conception de contrôleurs.Dans la deuxième partie, nous avons présenté une approche de conception conjointe intégrant les résultats obtenus dans les deux domaines: la synthèse et la conception des algorithmes de contrôle et l’ordonnancement de tâches temps réel qui partagent des ressources limitées. La technique proposée repose sur le changement de la période d’activation de l’algorithme de contrôle, et en conséquence le modèle du système devient un modèle échantillonné à taux variable. Les résultats proposés, en considérant l’algèbre de Lie des matrices d’évolution, permettent de calculer des contrôleurs adaptifs aux périodes qui stabilisent tous le système
The main contributions of this thesis are related to the analysis, synthesis and design of control systems sharing communication and computational resources. The research focuses on control systems where the feedback loops are closed over communication networks which transmit the information provided to its nodes by sensors, actuators and controllers. The shared resource in this scenario is the network. Some of the results are valid when the resource is a processor locally placed respect to several controller executing their algorithms on it. In any of the preceding scenarios, the control loops must contend for the shared resource. The limited capacity of the resource can cause delays and packet losses when information is transmitted. These effects can degrade the control system performance and even destabilize it.The first part of this thesis contributes to the performance analysis of specific classes of systems and to the design of robust controllers for network characteristics modeled by Quality of Service parameters. A series of methods to assist the control systems engineer are provided.In the second part, a contribution to the CoDesign approach is made via the integration of control system synthesis and design techniques with rules allowing to define the communication policy to manage real-time tasks sharing a limited resource. Putting in correspondence a scheduling of instances of the controller tasks with their sampling periods, the proposed policy results in discrete-time varying systems. The stabilization problem of these systems is solved with methods based on the solvability of Lie-algebras. Specifically, the proposed methodology provides adaptive controllers

Analyse des systemes non lineaires et leur commande par bouclage dynamique par approches structurelle et algebrique differentielles. Etude de l'inversion des systemes non lineaires a partir de notions algebriques et obtention de facon algorithmique de la structure a l'infini pour une classe generales de systemes non lineaires d'ou l'extension d'une liste d'invariants definie dans le cas de systemes lineaires invariants

The new Field Programmable Gate Array (FPGA) technologies and their structures have opened up new approaches to logic design and synthesis. The main feature of an FPGA is an array of logic blocks surrounded by a programmable interconnection structure. Cellular FPGAs are a special class of FPGAs which are distinguished by their fine granularity and their emphasis on local cell interconnects. While these characteristics call for specialized synthesis tools, the availability of logic gates other than Boolean AND, OR and NOT in these architectures opens up new possibilities for synthesis. Among the possible realizations of Boolean functions, XOR logic is shown to be more compact than AND/OR and also highly testable. In this dissertation, the concept of structural regularity and the advantages of XOR logic are used to investigate various synthesis approaches to cellular FPGAs, which up to now have been mostly nonexistent. Universal XOR Canonical Forms, Two-level AND/XOR, restricted factorization, as well as various Directed Acyclic Graph structures are among the proposed approaches. In addition, a new comprehensive methodology for the investigation of all possible XOR canonical forms is introduced. Additionally, a new compact class of XOR-based Decision Diagrams for the representation of Boolean functions, called Kronecker Functional Decision Diagrams (KFDD), is presented. It is shown that for the standard, hard, benchmark examples, KFDDs are on average 35% more compact than Binary Decision Diagrams, with some reductions of up to 75% being observed.

Doutoramento em Matemática
Nesta tese de Doutoramento desenvolve-se principalmente uma abordagem algébrica à teoria de sistemas de controlo não lineares. No entanto, outros tópicos são também estudados. Os tópicos tratados são os seguidamente enunciados: fórmulas para sistemas de controlo sobre álgebras de Lie livres, estabilidade de um sistema de corpos rolantes, algoritmos para aritmética digital, e equações integrais de Fredholm não lineares. No primeiro e principal tópico estudam-se representações para as soluções de sistemas de controlo lineares no controlo. As suas trajetórias são representadas pelas chamadas séries de Chen. Estuda-se a representação formal destas séries através da introdução de várias álgebras não associativas e técnicas específicas de álgebras de Lie livres. Sistemas de coordenadas para estes sistemas são estudados, nomeadamente, coordenadas de primeiro tipo e de segundo tipo. Apresenta-se uma demonstração alternativa para as coordenadas de segundo tipo e obtêm-se expressões explícitas para as coordenadas de primeiro tipo. Estas últimas estão intimamente ligadas ao logaritmo da série de Chen que, por sua vez, tem fortes relações com uma fórmula designada na literatura por “continuous Baker-Campbell- Hausdorff formula”. São ainda apresentadas aplicações à teoria de funções simétricas não comutativas. É, por fim, caracterizado o mapa de monodromia de um campo de vectores não linear e periódico no tempo em relação a uma truncatura do logaritmo de Chen. No segundo tópico é estudada a estabilizabilidade de um sistema de quaisquer dois corpos que rolem um sobre o outro sem deslizar ou torcer. Constroem-se controlos fechados e dependentes do tempo que tornam a origem do sistema de dois corpos num sistema localmente assimptoticamente estável. Vários exemplos e algumas implementações em Maple°c são discutidos. No terceiro tópico, em apêndice, constroem-se algoritmos para calcular o valor de várias funções fundamentais na aritmética digital, sendo possível a sua implementação em microprocessadores. São também obtidos os seus domínios de convergência. No último tópico, também em apêndice, demonstra-se a existência e unicidade de solução para uma classe de equações integrais não lineares com atraso. O atraso tem um carácter funcional, mostrando-se ainda a diferenciabilidade no sentido de Fréchet da solução em relação à função de atraso.
In this PhD thesis several subjects are studied regarding the following topics: formulas for nonlinear control systems on free Lie algebras, stabilizability of nonlinear control systems, digital arithmetic algorithms, and nonlinear Fredholm integral equations with delay. The first and principal topic is mainly related with a problem known as the continuous Baker-Campbell-Hausdorff exponents. We propose a calculus to deal with formal nonautonomous ordinary differential equations evolving on the algebra of formal series defined on an alphabet. We introduce and connect several (non)associative algebras as Lie, shuffle, zinbiel, pre-zinbiel, chronological (pre-Lie), pre-chronological, dendriform, D-I, and I-D. Most of those notions were also introduced into the universal enveloping algebra of a free Lie algebra. We study Chen series and iterated integrals by relating them with nonlinear control systems linear in control. At the heart of all the theory of Chen series resides a zinbiel and shuffle homomorphism that allows us to construct a purely formal representation of Chen series on algebras of words. It is also given a pre-zinbiel representation of the chronological exponential, introduced by A.Agrachev and R.Gamkrelidze on the context of a tool to deal with nonlinear nonautonomous ordinary differential equations over a manifold, the so-called chronological calculus. An extensive description of that calculus is made, collecting some fragmented results on several publications. It is a fundamental tool of study along the thesis. We also present an alternative demonstration of the result of H.Sussmann about coordinates of second kind using the mentioned tools. This simple and comprehensive proof shows that coordinates of second kind are exactly the image of elements of the dual basis of a Hall basis, under the above discussed homomorphism. We obtain explicit expressions for the logarithm of Chen series and the respective coordinates of first kind, by defining several operations on a forest of leaf-labelled trees. It is the same as saying that we have an explicit formula for the functional coefficients of the Lie brackets on a continuous Baker-Campbell-Hausdorff-Dynkin formula when a Hall basis is used. We apply those formulas to relate some noncommutative symmetric functions, and we also connect the monodromy map of a time-periodic nonlinear vector field with a truncation of the Chen logarithm. On the second topic, we study any system of two bodies rolling one over the other without twisting or slipping. By using the Chen logarithm expressions, the monodromy map of a flow and Lyapunov functions, we construct time-variant controls that turn the origin of a control system linear in control into a locally asymptotically stable equilibrium point. Stabilizers for control systems whose vector fields generate a nilpotent Lie algebra with degree of nilpotency · 3 are also given. Some examples are presented and Maple°c were implemented. The third topic, on appendix, concerns the construction of efficient algorithms for Digital Arithmetic, potentially for the implementation in microprocessors. The algorithms are intended for the computation of several functions as the division, square root, sines, cosines, exponential, logarithm, etc. By using redundant number representations and methods of Lyapunov stability for discrete dynamical systems, we obtain several algorithms (that can be glued together into an algorithm for parallel execution) having the same core and selection scheme in each iteration. We also prove their domains of convergence and discuss possible extensions. The last topic, also on appendix, studies the set of solutions of a class of nonlinear Fredholm integral equations with general delay. The delay is of functional character modelled by a continuous lag function. We ensure existence and uniqueness of a continuous (positive) solution of such equation. Moreover, under additional conditions, it is obtained the Fr´echet differentiability of the solution with respect to the lag function.

Thermodynamic processes in finite time are in general irreversible. But there are chances to avoid irreversibility. For instance, there are canonical ensembles of special quantum systems with a given probability distribution describing the likelihood to find the system at time t=0 in a particular state with energy E_i(0), which can be controlled in a specific way, such that the initial probability distribution is recovered at the end of the process (t=T), but the state energies did change, hence E_i(0) is not equal to E_i(T). This allows to change thermodynamic quantities (expectation values) adiabatically, reversibly and in finite time. Such special processes are called Shortcuts to Adiabaticity. The presented thesis analyzes the origin of these shortcuts utilizing special Hamiltonian systems with dynamical algebra. Their main feature is to provide canonical invariance, which means a canonical ensemble stays canonical under Hamiltonian dynamics. This invariance carried by the dynamical algebra will be discussed using Lie group theory. In addition, the persistence of the dynamical algebra with respect to calculating expectation values will be deduced. This allows to benefit from all intrinsic symmetries within the discussion of ensemble trajectories. In consequence, these trajectories will evolve under Hamiltonian dynamics on a specific manifold given by the so-called Casimir companion. In addition, the deformation of this manifold due to non-Hamiltonian (dissipative) dynamics will be discussed, which allows to present a framework for modeling irreversible processes based on Hamiltonian systems with dynamical algebra. An application of this framework based on the parametric harmonic oscillator will be presented by determining time-optimal controls for transitions between two equilibrium as well as between non-equilibrium and equilibrium states. The latter one will lead to time-optimal equilibration strategies for a statistical ensemble of parametric harmonic oscillators
Thermodynamische Prozesse in endlicher Zeit sind im Allgemeinen irreversibel. Es gibt jedoch Möglichkeiten, diese Irreversibilität zu umgehen. Ein kanonisches Ensemble eines speziellen quantenmechanischen Systems kann zum Beispiel auf eine ganz spezielle Art und Weise gesteuert werden, sodass nach endlicher Zeit T wieder eine kanonische Besetzungverteilung hergestellt ist, sich aber dennoch die Energie des Systems geändert hat (E(0) ungleich E(T)). Solche Prozesse erlauben das Ändern thermodynamischer Größen (Ensemblemittelwerte) der erwähnten speziellen Systeme in endlicher Zeit und auf eine adiabatische und reversible Art. Man nennt diese Art von speziellen Prozessen Shortcuts to Adiabaticity und die speziellen Systeme hamiltonsche Systeme mit dynamischer Algebra. Die vorliegende Dissertation hat zum Ziel den Ursprung dieser Shortcuts to Adiabaticity zu analysieren und eine Methodik zu entwickeln, die es erlaubt irreversible thermodynamische Prozesse adequat mittels dieser speziellen Systeme zu modellieren. Dazu wird deren besondere Eigenschaft ausgenutzt, die kanonische Invarianz, d.h. ein kanonisches Ensemble bleibt kanonisch bezüglich hamiltonscher Dynamik. Der Ursprung dieser Invarianz liegt in der dynamischen Algebra, die mit Hilfe der Theorie der Lie-Gruppen näher betrachtet wird. Dies erlaubt, eine weitere besondere Eigenschaft abzuleiten: Die Ensemblemittelwerte unterliegen ebenfalls den Symmetrien, die die dynamische Algebra widerspiegelt. Bei näherer Betrachtung befinden sich alle Trajektorien der Ensemblemittelwerte auf einer Mannigfaltigkeit, die durch den sogenannten Casimir Companion beschrieben wird. Darüber hinaus wird nicht-hamiltonsche/dissipative Dynamik betrachtet, welche zu einer Deformation der Mannigfaltigkeit führt. Abschließend wird eine Zusammenfassung der grundlegenden Methodik zur Modellierung irreversibler Prozesse mittels hamiltonscher Systeme mit dynamischer Algebra gegeben. Zum besseren Verständnis wird ein ausführliches Anwendungsbeispiel dieser Methodik präsentiert, in dem die zeitoptimale Steuerung eines Ensembles des harmonischen Oszillators zwischen zwei Gleichgewichtszuständen sowie zwischen Gleichgewichts- und Nichtgleichgewichtszuständen abgeleitet wird

De nombreux systèmes de production peuvent être modélisés et analysés à l’aide de graphes d’événements temporisés (GET). Les GET forment une classe de systèmes à événements discrets temporisés (SEDT), dont la dynamique est définie uniquement par des phénomènes de synchronisation et de saturation. Un avantage majeur des GET par rapport à d’autres classes de SEDT est qu’ils admettent, sous certaines conditions, un modèle linéaire dans des espaces algébriques particuliers : les dioïdes. Ceci a conduit au développement d’une théorie des systèmes linéaires dans les dioïdes, grâce à laquelle de nombreux concepts de l’automatique classique ont été adaptés aux GET. Par exemple, l’algèbre (max,+) (i.e., le dioïde basé sur les opérations (max,+)) offre des techniques élégantes pour l’analyse et le contrôle de GET. Cependant, les conditions nécessaires pour modéliser un système à événements discrets par un GET sont très restrictives. Pour élargir la classe de systèmes concernés, deux extensions principales ont été développées. D’une part, les GET valués ont été introduits pour décrire des phénomènes d’assemblage et de séparation dans les systèmes de production. Cette extension se traduit par l’association de coefficients entiers aux arrêtes d’un graphe d’événements. Contrairement aux GET, ces systèmes ne sont pas invariants par rapport aux événements et ne peuvent donc pas être décrits par des équations linéaires dans l’algèbre (max,+). D’autre part, la synchronisation partielle (PS) a été introduite pour modéliser des systèmes dans lesquels certains événements ne peuvent se produire que pendant des intervalles prédéfinis. Par exemple, dans une intersection réglée par un feu tricolore, une voiture peut traverser l’intersection lorsque le feu est vert. Contrairement aux GET, ces systèmes ne sont pas invariants dans le domaine temporel et ne peuvent donc pas être décrits par des équations linéaires dans l’algèbre (max,+). Dans cette thèse, une modélisation des GET valués et des GET avec PS dans des dioïdes adaptés est présentée. A l’aide de ces dioïdes, une décompostion pour les GET valués (resp. GET avec PS) en un GET et une partie non-invariante dans le domaine des événements (resp. dans le domaine temporel) est introduite. Sous certaines conditions, la partie invariante est invertible. Dans ce cas, les modèles et contrôleurs pour le GET valué ou le GET sous PS peuvent être directement dérivés des modèles et contrôleurs obtenus pour le GET associé
Various kinds of manufacturing systems can be modeled and analyzed by Timed EventGraphs (TEGs). These TEGs are a particular class of timed Discrete Event Systems (DESs), whose dynamic behavior is characterized only by synchronization and saturation phenomena. A major advantage of TEGs over many other timed DES models is that their earliest behavior can be described by linear equations in some tropical algebra structures called dioids. This has led to a broad theory for linear systems over dioids where many concepts of standard systems theory were introduced for TEGs. For instance, with the (max,+)-algebra linear state-space models for TEGs were established. These linear models provide an elegant way to do performance evaluation for TEGs. Moreover, based on transfer functions in dioids severalcontrol problems for TEGs were addressed. However, the properties of TEGs, and thus the systems which can be described by TEGs, are limited. To enrich these properties, two main extensions for TEGs were introduced. First,Weighted Timed Event Graphs (WTEGs) which, in contrast to ordinary TEGs, exhibit event-variant behaviors. InWTEGs integer weights are considered on the arcs whereas TEGs are restricted to unitary weights. For instance, these integer weights make it straightforward to model a cutting process in a production line. Second, a new kind of synchronization called partial synchronization (PS) was introduced forTEGs. PS is useful to model systems where specific events can only occur in a particular time window. For example, consider a crossroad controlled by a traffic light: the green phase of the traffic light provides a time window in which a vehicle is allowed to cross. Clearly, PS leads to time-variant behavior. As a consequence, WTEGs and TEGs under PS are not (max,+)-linear anymore.In this thesis, WTEGs and TEGs under PS are studied in a dioid structure. Based on these dioid models forWTEGs a decomposition of the dynamic behavior into an event-variant and an event-invariant part is proposed. Under some assumptions, it is shown that the eventvariant part is invertible. Hence, based on this model, optimal control and model reference control, which are well known for ordinary TEGs, are generalized to WTEGs. Similarly, a decomposition model is introduced for TEGs under PS in which the dynamic behavior is decomposed into a time-variant and time-invariant part. Again, under some assumptions, it is shown that the time-variant part is invertible. Subsequently, optimal control, as well as model reference control for TEGs under PS is addressed

In this thesis, we develop preconditioned iterative methods for the solution of matrix systems arising from PDE-constrained optimization problems. In order to do this, we exploit saddle point theory, as this is the form of the matrix systems we wish to solve. We utilize well-known results on saddle point systems to motivate preconditioners based on effective approximations of the (1,1)-block and Schur complement of the matrices involved. These preconditioners are used in conjunction with suitable iterative solvers, which include MINRES, non-standard Conjugate Gradients, GMRES and BiCG. The solvers we use are selected based on the particular problem and preconditioning strategy employed. We consider the numerical solution of a range of PDE-constrained optimization problems, namely the distributed control, Neumann boundary control and subdomain control of Poisson's equation, convection-diffusion control, Stokes and Navier-Stokes control, the optimal control of the heat equation, and the optimal control of reaction-diffusion problems arising in chemical processes. Each of these problems has a special structure which we make use of when developing our preconditioners, and specific techniques and approximations are required for each problem. In each case, we motivate and derive our preconditioners, obtain eigenvalue bounds for the preconditioners where relevant, and demonstrate the effectiveness of our strategies through numerical experiments. The goal throughout this work is for our iterative solvers to be feasible and reliable, but also robust with respect to the parameters involved in the problems we consider.

Le travail présenté s'inscrit dans le contexte de la théorie des systèmes linéaires dans les dioïdes. La motivation initiale de cette étude a été de contribuer à l'analyse et la commande de systèmes linéaires dans max-plus en utilisant spécifiquement une approche géométrique. La contribution de cette thèse est centrée sur deux problèmes. La première partie est dédiée à l'étude de la relation entre les notions d'invariance contrôlée et d'invariance contrôlée par retour d'état dynamique dans un semi-anneau. Cette relation permet de montrer l'équivalence de ces deux notions. La deuxième partie concerne un problème original dans la théorie des systèmes linéaires dans max-plus, il s'agit de la synthèse d'une loi de commande par retour d'état, qui permette de satisfaire un ensemble de spécifications exprimées sous la forme de restrictions sur l'état du système, avec une approche géométrique. Il s'agit plus précisément de commander des systèmes à événements discrets décrits par un modèle linéaire dans max-plus. Nous définissons et caractérisons l'ensemble des conditions initiales admissibles, lesquelles sont à l'origine de solutions non décroissantes. Les restrictions temporelles imposées à l'espace d'état du système sont décrites par le semi-module défini par l'image de l'étoile de Kleene de la matrice associée aux restrictions temporelles. Les propriétés géométriques de ce semi-module permettant de garantir que l'évolution du système en boucle fermée satisfasse les restrictions sont étudiées. Des conditions suffisantes concernant l'existence d'une loi de commande causale par retour d'état statique sont présentées. Le calcul des lois de commande causales est également présenté. Pour illustrer l'application de cette approche, deux problèmes de commande sont présentés
This work is in the context of the theory of linear Systems in the dioids. The initial motivation of this study was to contribute to the analysis and control of max-plus linear systems, specifically using a geometric approach. The contribution of this thesis focuses on two issues. The first part is dedicated to study of the relationship between the concepts of controlled invariance and dynamic state feedback controlled invariance in a semi-ring. This relationship allows us to show the equivalence of these two concepts. The second part relates to a new problem in the theory of max-plus linear systems, it is the synthesis, with a geometric approach, of a static state feedback control law, in order to satisfy a set of specifications that apply to the state space of the system. This is specifically to control of discrete event systems described by a linear model in max-plus. We define and characterize the set of admissible initial conditions, which are the cause of non-decreasing solutions. Temporal restrictions on the system state space are described by the semi-module defined by the image of the Kleene star of the matrix associated with time restrictions. The geometric properties of this semi-module are studied. Sufficient conditions for the existence of a causal control law by static feedback are presented. Calculating causal control laws is also presented. To illustrate the application of this approach, two control problems are presented

La qualité de la dérivation numérique en ligne d'un signal bruité joue un rôle primordial dans diverses applications touchant différents domaines (1D et 2D). Dans le travail présenté, nous nous sommes intéressés aux dérivateurs type "signal". Nos travaux s'inscrivent dans ce contexte et abordent des aspects théoriques et applicatifs. Du point de vue théorique, une étude de deux familles de dérivateurs a été effectuée. En premier lieu, nous avons étudié les dérivateurs de nature algébrique [travaux du projet INRIA-Non-A] et en deuxième lieu les différentiateurs modes glissants d'ordre supérieur [travaux de Levant]. Cette étude nous a permis de décrire les points forts et les points faibles de chacune de ces approches. Suite à cette étude, une nouvelle version des dérivateurs modes glissants a été proposée. Du point de vue applicatif, une première application dans le domaine 1D est présentée et qui consiste à utiliser le dérivateur proposé pour l'identification paramétrique en ligne d'un robot SCARA à deux axes. La méthode d'identification est validée expérimentalement par comparaison des résultats obtenus avec ceux donnés par d'autres méthodes (méthode en ligne et hors ligne). La deuxième application touche au domaine 2D. Dans ce cas, nous proposons l'application du différentiateur pour la détection des contours dans une image. Pour la validation des résultats, différents essais sont réalisés pour différents types de bruits. Une étude comparative avec des méthodes classiques est effectuée. Pour pouvoir tester l'efficacité du différentiateur dans des boucles de commande nous nous sommes intéressés aux systèmes électrohydrauliques. Le système physique étudié est un servo-vérin électrohydraulique à hautes performances ayant différents modes de fonctionnement. Après une succession d'hypothèses effectuées sur le modèle de simulation, des modèles de commande sont fournis. La dernière partie de ces travaux est dédiée à la synthèse de stratégie de commande via la technique du Backstepping pour le positionnement de l'axe électrohydraulique afin de tester l'efficacité l'algorithme proposé dans des boucles de commande. Comme base de comparaison, la version classique des algorithmes modes glissants est aussi utilisée.

The MSc Thesis deals with real-time modelling, especially with the control circuits, where we focus on the regulation circuit with auxiliary model of the regulated system. The introduction contains the Theory of Controls and regulation. The work is divided into two basic parts. The first part deals with the problems from the theoretical point of view. The theoretical part is concerned with the general systems. Various definitions of systems, their partition and mathematical methods for their description are mentioned after introducing. A description the regulation circuits, their partition, structures and types follows. There are shown the particular regulators and different types of regulated systems in details. Mathematical methods applied to solve the differential equations that describe the models of the systems are summarized in the second part of the work. The end of this chapter is concerned with the TKSL system. In the practical part I have produced Microsoft Power Point program which summarizes the results of experiments, made in the TKSL and TKSL/C system. The output of simulation was transformed into the graphs in the format of Microsoft Excel. The last part deals with suggestion and implementation simulators of control circuits.

碩士
國立中央大學
資訊及電子工程研究所
82
In this thesis,an algebraic approach to robust H-infinity control system is proposed.Firstly,a system with time-varying norm-bounded parameter uncertainty is considered.We also derive the dynamic output feedback controller based on the observer- estimator in order to stabilize the linear continuous-time sys- tems with time-varying norm-bounded uncertainties and,at the same time,to achieve the H-infinity disturbance attenuation under a prescribed level r,as well. Next,the discrete-time systems with time-varying norm-bounded parameter uncertainties is considered.We also derive a suffici- ent condition for obtaining the state feedback controller sati- sfying the robust stability and H-infinity constraint. At last,we consider the system with real-parameter uncertain- ties and unmodeled dynamics.We have another H-2 performance requirement in addition to the stability and H-infinity constr- aint for such system.Here,we derive a system of four coupled modified Riccation equations for obtaining the compensator satisfying our design criteria.

Sliding mode control has recently proved to be a highly effective method for state space modeling of differential-algebraic equation systems (DAEs). Sliding control realizations have great potential for simulation since they allow more computationally efficient robust modeling approximations to be constructed for DAE systems. However, efficient discretization of such methods poses a significant problem due to the well known chattering phenomena that often occurs due to limited computational bandwidth. While some errors are inevitable due to limited bandwidth, the chattering phenomenon can be reduced by minimizing the frequency at which the system crosses the sliding surface. In this work, we find relations between controller parameters, error bounds, and the crossing frequency. They are then used to synthesize efficient discretized sliding mode realizations that optimize crossing frequency and the associated controller sampling period. Together, these results form an efficient discretized bounded sliding mode control approach for simulation of differential-algebraic systems.

Differential-algebraic Equation (DAE) systems present numerous difficulties in distributed simulation and control systems. The main problem is that most existing methods require an explicit state space model without algebraic constraints. One approach to address this problem is to reformulate the DAE system into an equivalent nonlinear control problem, in which the algebraic constraints are replaced by appropriate sliding manifolds. However, previous approaches based on this method are centralized leading to a great deal of computation and communication in distributed environments associated with inversion of the associated input decoupling matrix. In this work, this problem is addressed through application of decentralized sliding mode control. Relationships are developed for stability and performance in the presence of the neglected coupling terms. Inversion of the decoupling matrix is performed on a node basis. This allows the systematic division of the system into nodes that are more efficient for distributed computation. The new approach is applied to simulation of deformable surfaces in a real-time distributed computing environment.

博士
國立成功大學
化學工程研究所
79
This dissertation is concerned with the problems of optimal parameter estima-tion and control for chemical processes described by nonlinear differential-algebraic equations. First, a differential-algebraic model is proposed for describing the dynamics of a liquid-liquid phase-transfer catalyzed batch reaction system. A two-stage optimal parameter estimation method combining Sobol systematic search and DSC-Powell sequential one is used to estimate model parameters. The reversible reaction between organic-phase benzyl chloride and aqueous-phase sodium bromide with tetrabutylammonia bromide as catalyst was carried out to verify the mathe-matical model. Simulation results reveal that the proposed model can adequately describe the dynamics and internal behavior of a liquid-liquid phase-transfer ct-alyzed reaction system. It is therefore convinced that the new differential-algebraic model would be helpful to the reactor design and optimal control of liquid-liquid phase-transfer catalyzed reaction systems. In solving a combined problem of optimal system parameter selection and con-trol for nonlinear differential-algebraic systems (DAS)involving various types of constraints, the control parametrization technique is used to approximate the orig-inal problem into a sequence of finite-dimensional parameter selection problems. Based on characterizing the Hamiltonian function and adjoint system associated with a DAS, and transforming different types of constrai ts into fund\ctionsls having the form of general cost functional, a unified gradient computation algorithm for the controls and system parameters is derived. This algorithm enables the result-ing approximate problems to be effectively solved by gradient-based optimization methods. In order to provide rigorous theoretical backgorund of this optimization scheme for DASs, its convergence properties are studied. The error bounds between the sub-optimal costs and the true optimal cost are derived. For the purpose of implementing controls with low-cost actuators, an opimal on-off control strategy is proposed for optimizing differential-algebraic process sys-tems. Inherently, the optimal control problem with the on-off control parametrization is a finite-dimensional parameter selection problem with the controller output levels and sequences of on-state duration times as parameters. A gradient algorithm for computing the gradients of cost functional and terminal constraints with respect to parameters are also derived. This gradient algorithm facilitates finding the solu-tion to the resulting optimal parameter selection problem by exiting gradient-based optimization methods. To illustrate the the proposed optimization schemes for DASs, the piecewise-constant control parametrization and the on-off control strategy are, respectively, applied to a singular clntrol problem of finding an optimal feeding policy for a fed-batch fermentation process governed by prodcut and substrate inhibited specific growth and product formation kinetics. The computed results are better than those given in literature. This shows that the proposed optimization methods are capable of solving singular control problems associated with a DAS. 本論文探討了同時伴有反應與平衡的化工程序系統之最佳參數合計與控制問題。首先 ，本論文提出描述液一液相間轉移催化反應系統之動態行為的數學模式，此模式同時 包含了用以描述油相中慢速反應之微分方程式，以及用以描述水相中快速解離平衡與 質量守恆的代數式。此外，吾人利用兩階段的參數估計法，來估測模式中之參數，克 服了複雜之系統參數不易由實驗直接求得的困難。為了驗證所提數學模式之適用性， 吾人進行了以氯芐（油相，溶劑為甲苯）與溴化鈉（水相），在以溴化四丁基銨為相 轉移觸媒的催化下，產生溴芐的可逆反應系統之實驗。在實驗與參數估計法的配合下 ，證實了吾人所提之微分代數式數學模式，可以準確地描述系統的動態行為，並可正 確的判斷四丁基銨鹽於兩相中是否處於萃取平衡，進而由此得知質傳之阻力可否予以 忽略。此外，也可很明顯的了解反應是單由質傳或單由反應所控制，抑或是先由質傳 控制而後轉為反應控制？另外，此類系統之鹽析效應，亦可由此模式來解釋。吾人深 信此數學模式，對於液─液相間轉移催化反應系統之反應器設計，以及最佳控制的研 究定有所助益。 其次，本論文更進一步探討了含有各種不同型式限制條件下的非線性微分代數程序系 統之最佳控制問題。吾人以控制參數化技術，由原來的控制問題產生一系列的近似問 題，而每一個所形成之近似問題，在本質上都是一有限空間的參數選擇問題，其解對 原最佳控制問題而言為一次最佳解。在處理限制條件方面，吾人將問題中之每一限制 式轉換為一個與目標函數相同型態之式子。再經由建立微分代數式系統之 Hamiltonian 函數與adjoint 系統，衍導出具有一致性的梯度計算公式。利用此梯度 式，吾人便可以一致的方法來計算目標函數以及各限制條件對控制參數化向量的梯度 值，因而可配合任一以梯度為主的最佳化方法，來解含有各種限制條件的非線性微分 代數程序系統之最佳控制問題。因為，以控制參數化技術來解非線性微分代數程序系 統的最佳控制問題是一新的研究，吾人更進一步地分析此一方法之收斂性，衍導了次 最佳目標函數值與最佳目標函數值間的差距範圍。 接下來，吾人更以經濟的觀點，探討了一低成本的最佳開關控制問題，此問題係決定 一序列控制器最佳開關時間，以及控制器最佳輸出準位。為了配合以梯度為主的最佳 化方法來解此最佳開關控制問題，吾人亦衍導出了一致性的梯度式，用來計算目標函 數與限制式對於最佳開啟時間與控制器輸出準位的梯度。此一實際且具經濟觀點的控 制策略，使吾人得以用低成本的致動器來實際控制微分代數程序系統。 最後，吾人將所提的最佳控制計算方法以及最佳開關控制策略，應用於一含有基質抑 制與產物抑制機制的資料批次醱酵程序系統。計算的結果較文獻中已發表之值為佳