\(H^ \infty\) - optimal control and related minimax design problems. A dynamic game approach (Q1188940)

The so-called `` \(H^ \infty\)-optimal control theory'' addresses the issue of worst-case controller design for linear plants subject to unknown additive disturbances, including problems of disturbance attenuation, model matching and tracking. In mid 1980's it was shown that a time-domain characterization of optimal or near-optimal controllers gives good results and is more encompassing, if compared to frequency- domain formulation used originally, since it allows one to formulate also finite-horizon and time-varying versions of the problem and thus also study the transient behavior. Among several different time-domain approaches to this class of worst- case controller design problems the one that uses the framework of dynamic (or differential) game theory was chosen by the authors as it proved to be most natural and simple. This is so because the original \(H^ \infty\)-optimal control problem (in its equivalent time-domain formulation) is in fact a minimax optimization problem, and hence a zero- sum game, where the controller can be viewed as the minimizing player and disturbance as the maximizing player. The book provides a self-contained exposition to the essential elements of linear-quadratic (LQ) zero-sum dynamic game theory, and shows how these tools can directly be used in solving the \(H^ \infty\)-optimal control problems in both continuous and discrete time, for finite and infinite horizons and under different perfect and imperfect state information patterns. In chapter 1 and 2 a brief history of the subject and formulation of the problems in view are given, basic notations and key results from zero-sum static and dynamic game theory are presented. Chapter 3 provides a complete solution to the basic discrete-time disturbance attenuation problems in both finite and infinite horizons, when the controller has access to the state with or without delay. The optimum controller is a memoryless feedback controller given in terms of the solution of a discrete-time Riccati equation or an algebraic Riccati equation. Chapter 4 develops the continuous time counterparts of the results of Chapter 2 under perfect state and sampled state information patterns. In contradiction with the discrete-time case a corresponding optimal controller may not exist. But there exist suboptimal linear feedback controllers ensuring attenuation levels arbitrary close to the optimal case. Chapter 5 considers the same problems as those in Chapter 4 in which the state variable is no longer available to the controller but only a disturbance-corrupted linear output is. The results are obtained for continuous, sampled and delayed imperfect state measurements. For the basic finite-horizon problems the characterization of a near-optimal controller is given in terms of the Riccati differential equation, one of which is a ``filtering'' Riccati equation. Chapter 6 studies the discrete time counterparts of the problems of Chapter 5. Chapter 7 discusses a class of minimax design problems in filtering and smoothing for linear time-varying systems related to the disturbance attenuation problem with imperfect measurements. Both discrete and continuous cases are considered. The book can be recommended to mathematicians specializing in control theory and dynamic (differential) games. It can be also incorporated into a second-level graduate course in a control curriculum as no background in game theory is required. The general framework adopted in the book should also make it possible for a reader to extend the present results to more general classes of systems such as nonlinear plants and distributed-parameter systems (some preliminary results on the former classes of problems are also presented in the book).