Zum Inhalt

Singular Linear-Quadratic Zero-Sum Differential Games and H∞ Control Problems

Regularization Approach

  • 2022
  • Buch

Über dieses Buch

Diese Monografie widmet sich der Analyse und Lösung von singulären Differentialspielen und singulären $H _ {inf} $-Kontrollproblemen sowohl in endlichen als auch in unendlichen Horizonten. Aufbauend auf den früheren Arbeiten der Autoren auf diesem Gebiet ist dieser neuartige Text der erste, der die oben erwähnten singulären Probleme mithilfe des Regularisierungsansatzes untersucht. Nach einer kurzen Einführung werden die Lösbarkeitsbedingungen für die regulären Differentialspiele und $H _ {inf} $-Kontrollprobleme vorgestellt. Im folgenden Kapitel lösen die Autoren das singuläre linear-quadratische Differentialspiel Finite-Horizont mit Hilfe der Regularisierungsmethode. Als nächstes wenden sie diese Methode auf die Lösung eines Typs mit unendlichem Horizont an. Die letzten beiden Kapitel widmen sich der Lösung der linearen-quadratischen $H _ {inf} $-Kontrollprobleme. Die Autoren verwenden theoretische und reale Beispiele, um die Ergebnisse und ihre Anwendbarkeit im gesamten Text zu veranschaulichen, und haben den Inhalt sorgfältig so in sich geschlossen wie möglich organisiert, so dass es möglich ist, jedes Kapitel einzeln oder nacheinander zu studieren. Jedes Kapitel enthält eine eigene Einleitung, eine Liste von Notizen, eine kurze Literaturübersicht zum Thema und eine entsprechende Bibliographie. Zur besseren Lesbarkeit werden detaillierte Beweise in separaten Unterabschnitten präsentiert. Einzelne linear-quadratische Nullsummenspiele und $H _ {inf} $Kontrollprobleme werden für Forscher und Ingenieure interessant sein, die in den Bereichen angewandte Mathematik, dynamisches Spiel, Regelungstechnik, Maschinenbau und Luft- und Raumfahrt, Elektrotechnik und Biologie arbeiten. Dieses Buch kann auch als nützliche Referenz für Doktoranden in diesen Bereichen dienen.

Inhaltsverzeichnis

  1. Frontmatter

  2. 1. Introduction

    Valery Y. Glizer, Oleg Kelis
    Abstract
    A singular differential game is the game which cannot be solved by the application of the first-order solvability conditions.
  3. 2. Examples of Singular Extremal Problems and Some Basic Notions

    Valery Y. Glizer, Oleg Kelis
    Abstract
    In this chapter, we present several examples of singular extremal problems. Some of these examples are academic ones.
  4. 3. Preliminaries

    Valery Y. Glizer, Oleg Kelis
    Abstract
    In this chapter, several problems, allowing first-order solvability conditions, are considered. Namely, we consider zero-sum linear-quadratic differential games in finite-horizon and infinite-horizon settings, as well as \(H_{\infty }\) control problems in finite-horizon and infinite-horizon settings.
  5. 4. Singular Finite-Horizon Zero-Sum Differential Game

    Valery Y. Glizer, Oleg Kelis
    Abstract
    In this chapter, a finite-horizon zero-sum linear-quadratic differential game, which cannot be solved by application of the first-order solvability conditions, is considered. Thus, this game is singular. Its singularity is due to the singularity of the weight matrix in the control cost of a minimizing player in the game’s functional. For this game, novel definitions of a saddle-point equilibrium (a saddle-point equilibrium sequence) and a game value are introduced. Regularization method is proposed for obtaining these saddle-point equilibrium sequence and game value. This method consists in an approximate replacement of the original singular game with an auxiliary regular finite-horizon zero-sum linear-quadratic differential game depending on a small positive parameter. Thus, the first-order solvability conditions are applicable for this new game. Asymptotic analysis (with respect to the small parameter) of the Riccati matrix differential equation, arising in these conditions, yields the solution (the saddle-point equilibrium sequence and the game value) to the original singular game.
  6. 5. Singular Infinite-Horizon Zero-Sum Differential Game

    Valery Y. Glizer, Oleg Kelis
    Abstract
    In this chapter, an infinite-horizon zero-sum linear-quadratic differential game, which cannot be solved by application of the first-order solvability conditions, is considered. This game is singular, and its singularity is due to the singularity of the weight matrix in the control cost of a minimizing player in the game’s functional. For this game, novel definitions of a saddle-point equilibrium (a saddle-point equilibrium sequence) and a game value are introduced. Regularization method is proposed for obtaining these saddle-point equilibrium sequence and game value. This method consists in an approximate replacement of the original singular game with an auxiliary regular infinite-horizon zero-sum linear-quadratic differential game depending on a small positive parameter. Thus, the first-order solvability conditions are applicable for this new game. Asymptotic analysis (with respect to the small parameter) of the Riccati matrix algebraic equation, arising in these conditions, yields the solution (the saddle-point equilibrium sequence and the game value) to the original singular game.
  7. 6. Singular Finite-Horizon Problem

    Valery Glizer, Oleg Kelis
    Abstract
    In this chapter, we consider a system consisting of an uncertain controlled linear time-dependent differential equation and a linear time-dependent output algebraic equation. For this system, a finite-horizon \(H_{\infty }\) problem is studied in the case where the rank of the coefficients’ matrix for the control in the output equation is smaller than the Euclidean dimension of this control. In this case, the solvability conditions, based on the game-theoretic matrix Riccati differential equation, are not applicable to the solution of the considered \(H_{\infty }\) problem meaning its singularity. To solve this \(H_{\infty }\) problem, a regularization method is proposed. Namely, the original problem is replaced approximately with a regular finite-horizon \(H_{\infty }\) problem depending on a small positive parameter. Thus, the first-order solvability conditions are applicable to this new problem. Asymptotic analysis (with respect to the small parameter) of the Riccati matrix differential equation, arising in these conditions, yields a controller solving the original singular \(H_{\infty }\) problem. Properties of this controller are studied.
  8. 7. Singular Infinite-Horizon Problem

    Valery Y. Glizer, Oleg Kelis
    Abstract
    In this chapter, we consider a system consisting of an uncertain controlled linear time-invariant differential equation and a linear time-invariant output algebraic equation. For this system, an infinite-horizon \(H_{\infty }\) problem is studied in the case where the rank of the coefficients’ matrix for the control in the output equation is smaller than the Euclidean dimension of this control. In this case, the solvability conditions, based on the game-theoretic matrix Riccati algebraic equation, are not applicable to the solution of the considered \(H_{\infty }\) problem meaning its singularity. To solve this \(H_{\infty }\) problem, a regularization method is proposed. Namely, the original problem is replaced approximately with a regular infinite-horizon \(H_{\infty }\) problem depending on a small positive parameter. Thus, the first-order solvability conditions are applicable to this new problem. Asymptotic analysis (with respect to the small parameter) of the Riccati matrix algebraic equation, arising in these conditions, yields a controller solving the original singular \(H_{\infty }\) problem. Properties of this controller are studied.
  9. Backmatter

Titel
Singular Linear-Quadratic Zero-Sum Differential Games and H∞ Control Problems
Verfasst von
Valery Y. Glizer
Oleg Kelis
Copyright-Jahr
2022
Electronic ISBN
978-3-031-07051-8
Print ISBN
978-3-031-07050-1
DOI
https://doi.org/10.1007/978-3-031-07051-8

Informationen zur Barrierefreiheit für dieses Buch folgen in Kürze. Wir arbeiten daran, sie so schnell wie möglich verfügbar zu machen. Vielen Dank für Ihre Geduld.

    Bildnachweise
    Schmalkalden/© Schmalkalden, NTT Data/© NTT Data, Verlagsgruppe Beltz/© Verlagsgruppe Beltz, EGYM Wellpass GmbH/© EGYM Wellpass GmbH, rku.it GmbH/© rku.it GmbH, zfm/© zfm, ibo Software GmbH/© ibo Software GmbH, Sovero/© Sovero, Axians Infoma GmbH/© Axians Infoma GmbH, genua GmbH/© genua GmbH, Prosoz Herten GmbH/© Prosoz Herten GmbH, Stormshield/© Stormshield, MACH AG/© MACH AG, OEDIV KG/© OEDIV KG, Rundstedt & Partner GmbH/© Rundstedt & Partner GmbH, Doxee AT GmbH/© Doxee AT GmbH , Governikus GmbH & Co. KG/© Governikus GmbH & Co. KG