No description available.

Dans la plupart des problèmes de synthèse, la loi de commande obtenue doit répondre simultanément à des critères réquentiels et temporels en vue de satisfaire un cahier des charges précis. Les derniers développements des techniques de synthèse Hinf de contrôleurs structurés permettent d'obtenir des lois de commande satisfaisant des critères fréquentiels multiples. En revanche, la synthèse de loi de commande satisfaisant une contrainte temporelle sur une sortie ou un état du système est plus complexe. Dans ce travail de thèse, la technique OIST est considérée pour ce type de contraintes. Elle consiste à saturer la sortie du contrôleur dès que la contrainte n'est plus vérifiée afin de restreindre l'ensemble des sorties admissibles. Initialement formulée pour les systèmes linéaires connus dont l'état est mesuré, la technique OIST peut être généralisée pour permettre de considérer des systèmes incertains. C'est l'extension OISTeR qui est proposée dans ce travail. Elle utilise les données d'un observateur par intervalles pour borner de manière garantie le vecteur d'état. La théorie des observateurs par intervalles a récemment fait l'objet de nombreux travaux. La méthode la plus rapide pour obtenir un observateur par intervalles d'un système donné est de considérer un système intermédiaire coopératif dans de nouvelles coordonnées. Une nouvelle technique de détermination de ces nouvelles coordonnées, intitulée SCorplO, est proposée dans ce mémoire. L'ensemble des techniques présentées est appliqué au contrôle d'un lanceur flexible durant son vol atmosphérique, en présence de rafales de vent et sous contrainte temporelle sur l'angle d'incidence. / Ln control design problems, both frequency- and time-domain requirements are usually considered such that the resulting control law satisfies the specifications. Novel non-smooth optimization techniques can be used to achieve multiple frequency-domain specifications over a family of linear models. However, enforcing time-domain constraints on a given output or state is more challenging Since translating them into frequency-domain requirements may be inaccurate. This motivates the study of an additional approach to the Hinf control design techniques. When time-domain constraints are satisfied, the nominal control law reduces to a controller satisfying the frequency-domain constraints. Upon violation of the ime-domain constraint, an additional tool named OIST is used to saturate the controller output so as to restrict the reachable set of the constrained system output. Stability guarantees are obtained for minimum phase systems. Further developments proposed therein allow the consideration of uncertain systems With incomplete state measurements. This is he OISTeR approach. The method uses certified bounds on the considered system state as provided by an interval observer. he theory of interval observers is well-established. ln the case of linear systems, the most common approach is to consider an intermediate cooperative system on which the interval observer can be built. The novel SCorplO design method proposed in this work is used to compute such cooperative representation. ln this thesis, the considered application is the atmospheric control of a flexible launch vehicle under a time-domain constraint on the angle of attack and in the presence of wind gusts.

In this thesis, an unmanned air vehicle (UAV) is used to develop a prototype base test platform for flight testing of new control algorithms and avionics for advanced UAV system development applications. A control system that holds the UAV at a fixed altitude above the ground is designed and flight tested. Only the longitudinal motion of the UAV is considered during the controller design, hence its lateral motions are controlled manually by a remote control unit from the ground. UAV&amp / #8217 / s altitude with respect to the mean sea level and position are obtained by an onboard global positioning system (GPS) and this information is transmitted to the ground computer via radio frequency (RF) communication modules. The altitude of the UAV above the ground is calculated by using the digital terrain elevation data (DTED). A controller is designed and its gains are tuned to maintain this flight altitude at a desired value by using the mathematical model developed to represent the longitudinal dynamics of the UAV. Input signals generated by the controller for elevator deflections are transmitted back to the UAV via RF communication modules to drive onboard servomotors to generate desired elevator deflections. All controller computations and RF communications are handled by a MATLAB&reg / based platform on a ground computer. UAV flight tests are carried out at two different autopilot modes / namely, mean sea level (MSL) altitude hold mode and above ground level (AGL) altitude hold mode. The developed platform worked properly during flight tests and proved to be reliable in almost every condition. Moreover, the designed controller system is demonstrated to be effective and it fulfills the requirements.

Orientador: Marcelo Carvalho Minhoto Teixeira / Banca: Edvaldo Assunção / Banca: Celso Pascoli Bottura / Resumo: Em alguns problemas práticos, por exemplo, no controle de vibrações de sistemas mecânicos utilizando acelerômetros como sensores, é mais fácil obter a da derivada do vetor de estado do que o vetor de estado. Esta dissertação mostra que plantas lineares e invariantes no tempo descritas pelas matrizes {A,B,C,D}, cuja saída é a derivada do vetor de estado, são não observáveis e não estabilizáveis com realimentação da saída, quando det(A)=0. A impossibilidade de rejeição de distúrbios constantes, somados à entrada da planta quando det(A) 6= 0, realimentando-se a saída com controladores estáticos, foi também demonstrada. Adicionalmente, são propostas novas técnicas de controle para uma classe de sistemas não lineares com incertezas variantes no tempo. Condições na forma de desigualdades matriciais lineares, em inglês Linear Matrix Inequalities (LMIs), para o projeto de sistemas de controle empregando realimentação derivativa estática da derivada do vetor de estado, que ao mesmo tempo estabiliza a planta e maximiza o limite de incerteza do termo não linear, são estabelecidas com base em funções quadráticas de Lyapunov. Posteriormente estendem-se as condições de estabilidade robusta para sistemas não lineares com incertezas variantes no tempo e incertezas politópicas, que podem também representar falhas estruturais. O projeto do controlador é realizado através de condições baseadas em LMIs que, quando factíveis, podem ser resolvidas facilmente utilizando técnicas de programação convexa. Essa metodologia permite a inclusão de restrições de desempenho no projeto, como por exemplo, na taxa de decaimento e na norma de ganho dos controladores, de modo a atender às restrições do projeto. São apresentadas análises e resultados considerando o sinal de controle nulo e empregando realimentação derivativa estática do vetor de estado... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: In some practical problems, for instance in the control of mechanical systems using accelerometers as sensors, it is easier to obtain the state-derivative vectors than the state vectors. This dissertation shows that (i) linear time-invariant plants given by the state-space model matrices {A,B,C,D} with output equal to the state-derivative vector are not observable and can not be stabilizable by using an output feedback if det(A) = 0 and (ii) the rejection of a constant disturbance added to the input of the aforementioned plants, considering det(A) 6= 0, and a static output feedback controller is not possible. The proposed results can be useful in the analysis and design of control systems with state-derivative feedback. Additionally, a new procedure for the control of a class of nonlinear systems with time-varying uncertainties using static statederivative feedback is proposed. Conditions based on Linear Matrix Inequalities (LMIs) for the design of robust controllers using state-derivative feedback, which simultaneously stabilizes the system and maximizes the uncertainty bound of the nonlinear term are proposed. Robust stability conditions for nonlinear systems with uncertainties and time-varying polytopic uncertainties, known as structural failures, are also presented. The controller design are based on LMIs which, when feasible, can be easily solved using convex programming techniques. This methodology allows the inclusion of performance constraints on the design, such as the decay rate and gain bounds in order to meet the design requirements. The dissertation considered cases where the control signal is equal to zero and with a state-derivative feedback. The elaboration of the theory for systems with n nonlinearities is illustrated through examples covering a single nonlinearity, with and without two polytopic uncertainties (and structural failures), and two nonlinearities... (Complete abstract click electronic access below) / Mestre

A novel controller termed as Experience Mapping based Prediction Controller (EMPC) is developed in this work. EMPC is developed utilizing the broad control concepts of human motor control (HMC). The concepts of HMC are utilized to develop the core concepts of EMPC for the control of ideal Type-1 LTI systems. The control accuracy of the developed concepts is studied and the mathematical stability criterion for the controller is developed. The applicability of EMPC for the control of real world problems is tested on a Permanent Magnet DC motor based position control system. 1. Novel learning methods are presented to form experience mapped knowledge-base (EMK) which is used for the creation of the forward and inverse models. 2. Control and Adaptation Techniques which overcome the presence of non idealities are developed using the inverse model. 3. Two separate techniques which utilize the forward model for improving the adaptation capabilities of EMPC are developed. 4. Two novel techniques are developed for the improvement of the tracking performance in terms of the accuracy and smoothness of tracking. These techniques are tested under various system conditions including large dynamic parameter changes on a simulation model and a practical setup. The performance of EMPC is compared against that of PID, MRAC and LQG controllers for all the proposed techniques and EMPC is found to perform significantly better under the various system conditions in terms of transient and steady state characteristics. Finally, the effectiveness of EMPC in stabilizing unstable systems using the concepts developed is tested on a practical Inverted Pendulum system. The problem of the simultaneous development of experiences and control of the system is addressed with the stabilizing problem. The proposed controller, EMPC provides an alternative approach for the existing control of systems without the requirement of an accurate system mathematical model. Its capability to learn by directly interacting with the system and adapt using experiences makes it an attractive alternative to other control techniques present in literature. Keywords: EMPC, Position Control, PMDC motors

Made available in DSpace on 2014-06-11T19:22:31Z (GMT). No. of bitstreams: 0 Previous issue date: 2011-02-28Bitstream added on 2014-06-13T18:49:28Z : No. of bitstreams: 1 moreira_mr_me_ilha.pdf: 1357345 bytes, checksum: 0cd0261b936a9aacf1f408322f1ec3be (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Em alguns problemas práticos, por exemplo, no controle de vibrações de sistemas mecânicos utilizando acelerômetros como sensores, é mais fácil obter a da derivada do vetor de estado do que o vetor de estado. Esta dissertação mostra que plantas lineares e invariantes no tempo descritas pelas matrizes {A,B,C,D}, cuja saída é a derivada do vetor de estado, são não observáveis e não estabilizáveis com realimentação da saída, quando det(A)=0. A impossibilidade de rejeição de distúrbios constantes, somados à entrada da planta quando det(A) 6= 0, realimentando-se a saída com controladores estáticos, foi também demonstrada. Adicionalmente, são propostas novas técnicas de controle para uma classe de sistemas não lineares com incertezas variantes no tempo. Condições na forma de desigualdades matriciais lineares, em inglês Linear Matrix Inequalities (LMIs), para o projeto de sistemas de controle empregando realimentação derivativa estática da derivada do vetor de estado, que ao mesmo tempo estabiliza a planta e maximiza o limite de incerteza do termo não linear, são estabelecidas com base em funções quadráticas de Lyapunov. Posteriormente estendem-se as condições de estabilidade robusta para sistemas não lineares com incertezas variantes no tempo e incertezas politópicas, que podem também representar falhas estruturais. O projeto do controlador é realizado através de condições baseadas em LMIs que, quando factíveis, podem ser resolvidas facilmente utilizando técnicas de programação convexa. Essa metodologia permite a inclusão de restrições de desempenho no projeto, como por exemplo, na taxa de decaimento e na norma de ganho dos controladores, de modo a atender às restrições do projeto. São apresentadas análises e resultados considerando o sinal de controle nulo e empregando realimentação derivativa estática do vetor de estado... / In some practical problems, for instance in the control of mechanical systems using accelerometers as sensors, it is easier to obtain the state-derivative vectors than the state vectors. This dissertation shows that (i) linear time-invariant plants given by the state-space model matrices {A,B,C,D} with output equal to the state-derivative vector are not observable and can not be stabilizable by using an output feedback if det(A) = 0 and (ii) the rejection of a constant disturbance added to the input of the aforementioned plants, considering det(A) 6= 0, and a static output feedback controller is not possible. The proposed results can be useful in the analysis and design of control systems with state-derivative feedback. Additionally, a new procedure for the control of a class of nonlinear systems with time-varying uncertainties using static statederivative feedback is proposed. Conditions based on Linear Matrix Inequalities (LMIs) for the design of robust controllers using state-derivative feedback, which simultaneously stabilizes the system and maximizes the uncertainty bound of the nonlinear term are proposed. Robust stability conditions for nonlinear systems with uncertainties and time-varying polytopic uncertainties, known as structural failures, are also presented. The controller design are based on LMIs which, when feasible, can be easily solved using convex programming techniques. This methodology allows the inclusion of performance constraints on the design, such as the decay rate and gain bounds in order to meet the design requirements. The dissertation considered cases where the control signal is equal to zero and with a state-derivative feedback. The elaboration of the theory for systems with n nonlinearities is illustrated through examples covering a single nonlinearity, with and without two polytopic uncertainties (and structural failures), and two nonlinearities... (Complete abstract click electronic access below)

Cette thèse est consacrée au développement des techniques d’estimation et de commande pour systèmes descripteurs non linéaires. Les développements sont centrés sur une famille particulière de systèmes descripteurs non linéaires avec une matrice descripteur de rang plein. Toutes les approches présentées utilisent un formalisme de modélisation du type Takagi-Sugeno (TS) pour représenter les modèles descripteurs non linéaires. Un objectif très important est de développer des conditions sous la forme d’inégalités matricielles linéaires (LMI, en anglais). Dans la littérature, les conditions pour l’estimation des modèles TS descripteurs s’écrivent sous forme d’inégalités matricielles bilinéaires (BMI, en anglais). En plus, à notre connaissance, il n’y pas de résultats dans la littérature concernant la commande/estimation pour les modèles TS descripteurs en temps discret (avec une matrice descripteur régulière non linéaire).Trois problèmes ont été examinés : commande par retour d’état, estimation de l’état et commande statique par retour de la sortie. Dans le cas continu, des conditions moins conservatives ont été développées pour la commande par retour d’état. Pour l’estimation d’état, des conditions LMI ont été obtenues (au lieu des usuelles BMI) en utilisant un différent vecteur d’erreur augmenté. Pour la commande statique par retour de la sortie, des conditions LMI sont proposées si une matrice auxiliaire est fixée. Pour le temps discret, des nouveaux résultats sous la forme LMI ont été développées pour la commande/estimation, comblant ainsi certains manques de la littérature. Des exemples ont été inclus pour montrer l’applicabilité de tous les résultats que nous avons obtenus et ainsi l’importance de garder la structure originale des descripteurs. / This thesis addresses the estimation and control for nonlinear descriptor systems. The developments are focused on a family of nonlinear descriptor models with a full-rank descriptor matrix. The proposed approaches are based on a Takagi-Sugeno (TS) descriptor representation of a given nonlinear descriptor model. This type of TS models is a generalization of the standard TS ones. One of the mains goals is to obtain conditions in terms of linear matrix inequalities (LMIs). In the existing literature, the observer design for TS descriptor models has led to bilinear matrix inequality (BMI) conditions. In addition, to the best of our knowledge, there are no results in the literature on controller/observer design for discrete-time TS descriptor models (with a non-constant and invertible descriptor matrix).Three problems have been addressed: state feedback controller design, observer design, and static output feedback controller design. LMI conditions have been obtained for both continuous and discrete-time TS descriptor models. In the continuous-time case, relaxed LMI conditions for the state feedback controller design have been achieved via parameterdependent LMI conditions. For the observer design, pure LMI conditions have been developed by using a different extended estimation error. For the static output feedback controller, LMI constraints can be obtained once an auxiliary matrix is fixed. In the discretetime case, results in the LMI form are provided for state/output feedback controller design and observer design; thus filling the gap in the literature. Several examples have been included to illustrate the applicability of the obtained results and the importance of keeping the original descriptor structure instead of computing a standard state-space.

No description available.

<p dir="ltr">Hydraulic actuation technology is one of the most common ways of transferring power in agricultural and construction machinery. As the system efficiency and fuel consumption of these machines gains more attention due to climate change, a new generation of high-efficient hydraulic architecture is in need. This dissertation presents a study for reducing energy loss in the hydraulic control system of agricultural tractors and their implements. The solution is referred to as a multi-pressure rail (MPR) system and provides power to the hydraulic functions following a pressure control logic, as opposed to the traditional flow control logic typical of hydraulic systems used in off-road vehicles.</p><p dir="ltr">The proposed hydraulic architecture and controller allows for elimination of redundant flow control valves in the state-of-the-art system, which cause excessive throttling losses leading to poor overall energy efficiency. Related work on MPR technology targets construction vehicles, where the MPR solution can allow energy recovery during overrunning loads and better engine management. This study alternatively addresses the case of agricultural applications where functions mostly operate under resistive load conditions with slow dynamics, which offers an opportunity to target throttle losses. The hydraulic architecture design starts with the choice of number of rails, then supply system and pressure selection and control valve set. Next, the controller is proposed. The controller contains two layers. The lower layer directly controls the command tracking, rotational speed, or pressure, for each subsystem. The higher layer, namely the supervisory controller, optimizes the rail pressure levels in real time to guarantee minimum overall throttling loss.</p><p dir="ltr">To prove the effectiveness of the hydraulic system architecture and controller design, a standalone test rig was conceived and used to validate a numerical simulation model of the MPR system and its control strategy. Particular focus is given to the dynamic behavior of the system during the switches of a function between different pressure rails, which needs to ensure reduced oscillations of the flow provided to each hydraulic function. Then, to demonstrate the ability on power saving in real working conditions, reference machines were chosen: a 435 hp hydraulic tractor powering a 16-row planter, for which operating features during typical drive cycles were available to the authors. Simulation models of the two reference machines were built and validated with in field experiments. A full MPR system model on the reference machines was constructed using the validated models. This full model was used to predict the reference tractor and planter hydraulic system performance and power consumption during typical drive cycles. The results show up to 52.4% total power reduction at the pump shaft, corresponding to 113.8% system efficiency gain.</p><p dir="ltr">The dissertation also laid out the planned activities to complete the study. The system controller will be generalized that could suit more equipment in addition to the reference machine in this study. In the meantime, the reference tractor and planter will be modified into the proposed MPR system for field testing. That includes new sensors, controllers, valves, etc. The field test is the final experimental validation for the proposed MPR system on the front of effectiveness and power saving in real working condition.</p>

Die Arbeit ist der regelungstheoretischen Betrachtung von mechanischen Systemen mit mehr Freiheitsgraden als Stellgrößen gewidmet. Dabei werden Aspekte aus den Teilgebieten Modellbildung, Systemanalyse, Steuerungsentwurf und Reglerentwurf behandelt. Den Ausgangspunkt bilden die aus dem Lagrange-Formalismus resultierenden Bewegungsgleichungen, für welche neben verschiedene partiell linearisierten Zustandsdarstellungen auch eine spezielle Byrnes-Isidori-Normalform eingeführt wird. Im Unterschied zu einer früher vorgeschlagenen ähnliche Normalform existiert diese "Lagrange-Byrnes-Isidori-Normalform" immer. Weiterhin wird die bedeutende Eigenschaft der differentiellen Flachheit im Zusammenhang mit mechanischen Systemen untersucht. Die bestehende Lücke zwischen den bekannten notwendigen und hinreichenden Flachheitsbedingungen bildet die Motivation zur Anpassung der Regelflächenbedingung auf mechanische Systeme in Lagrange-Byrnes-Isidori-Normalform. Parallel dazu wird die Flachheitsanalyse auf Basis des sogenannten Variationssystems betrachtet. Dabei handelt es sich um ein System von 1-Formen, die durch Anwendung der äußeren Ableitung auf die impliziten Systemgleichungen entstehen. Äquivalent dazu können auch die in einer rechteckigen Polynommatrix bezüglich des Zeitableitungsoperators zusammengefassten Koeffizienten der Basisformen untersucht werden. Die Flachheit eines Systems ist nun gerade äquivalent zur Existenz einer unimodularen Vervollständigung dieser Matrix, welche zudem noch eine bestimmte Integrabilitätsbedingung erfüllen muss. Durch Anwendung des Satzes von Frobenius können aus diesen in der bisherigen Formulierung nur schwer überprüfbaren Bedingungen deutlich einfachere hergeleitet werden. Für den Eingrößenfall ergibt sich dadurch eine erheblich Verringerung des Rechenaufwandes im Vergleich zum Referenzansatz. Im Mehrgrößenfall ist die Situation komplizierter: Durch das Fallenlassen der Unimodularitätsforderung und die Ausnutzung der speziellen Struktur mechanischer Systeme erhält man eine neue notwendige Bedingung für Flachheit, welche sich in endlich vielen Schritten auswerten lässt. Allerdings konnte mit dieser die vermutete Nichtflachheit für die untersuchten mechanischen Beispielsysteme nicht nachgewiesen werden. Einen weiteren Untersuchungsgegenstand bildet das Konzept der Konfigurationsflachheit. Für diese Eigenschaft ist gefordert, dass ein flacher Ausgang existieren muss, der nur von den Konfigurationskoordinaten abhängt. Basierend auf theoretischen Überlungen und dem Fehlen von Gegenbeispielen wird die Hypothese aufgestellt, dass für konservative mechanische Systeme Flachheit und Konfigurationsflachheit äquivalent sind. Für lineare mechanische Systeme kann diese Hypothese mit Hilfe der Kronecker-Normalform von Matrizenscharen verifiziert werden. Bezüglich des Entwurfs von Solltrajektorien werden neben der Darstellung bekannter Verfahren für lineare und für flache Systeme zwei weitere Ansätze genauer diskutiert. Der erste basiert auf der numerischen Lösung des aus dem Steuerungsentwurf resultierenden Randwertproblems. Dazu wird ein angepasstes Kollokationsverfahren konstruiert, welches die Elimination von Systemgrößen durch die explizite Berücksichtigung von Integratorketten ermöglicht, die bei partiell linearisierten Systemen stets auftreten. Unter bestimmten Bedingungen bewirkt dies eine erhebliche Reduktion der Rechenzeit. Der zweite Ansatz betrachtet die Überführung zwischen zwei Ruhelagen und beruht auf der Zeitumkehrsymmetrie, die alle konservativen mechanischen Systeme aufweisen. Er besteht aus mehreren Schritten: Zunächst wird für beide Ruhelagen eine Rückführung mit möglichst großem Attraktivitätsgebiet entworfen. Danach wird das System simulativ ausgehend von der Zielruhelage in der Startruhelage stabilisiert. Die so erhaltene Eingangstrajektorie kann dann bezüglich der Zeit invertiert werden, um das System aus der Startruhelage in die Nähe der Zielruhelage zu überführen, wo schließlich der entsprechende Regler aktiviert wird. In praktischen Realisierungen von unteraktuierten Regelungssystemen treten auf Grund von Effekten wie trockener Reibung und Getriebespiel oft Dauerschwingungen mit schwer vorhersagbaren und beeinflussbaren Parametern auf. Als Alternative zur klassischen Stabilisierung einer (theoretischen) Ruhelage wird deshalb eine Rückführung hergeleitet, welche für ein gegebenes lineares System einen stabilen Grenzzyklus mit vorgebbarer Frequenz und Amplitude asymptotisch stabilisiert.