Robust H infinity suboptimal and guaranteed cost state feedbacks as solutions to linear-quadratic dynamic games under uncertainty (Q4485355)

The author deals with the linear-quadratic dynamic games in the case where the system equation contains unknown norm-bounded time-varying functions. He constructs two auxiliary dynamic games with known equations such that a minimax strategy for one of them and a maximin strategy for the other are the robust minimax and maximin strategies for the original dynamic game. The problem of robust \(H_{\infty}\) performance was considered by \textit{L. Xie} and \textit{C. E. de Souza} [IEEE Trans. Autom. Control 37, No. 8, pt. 1, 1188-1191 (1992; Zbl 0764.93027)] and \textit{L. Yuan, L. E. K. Achenie} and \textit{W. Jiang} [Syst. Control Lett. 27, No. 4, 199-208 (1996; Zbl 0875.93108)] for continuous-time and discrete-time systems provided that there is a matching in the structure of the uncertain state and input matrices and there is no uncertainty in the disturbance matrix. They show that the problem of robust \(H_{\infty}\) control can be reduced to an \(H_{\infty}\) synthesis problem of a system which is independent of the uncertainty in the system. In this paper the author extends the robust \(H_{\infty}\) control and the guaranteed cost controllers to the wider class of uncertain discrete-time systems and shows that the rejection of the matching condition results in a two-parameter Riccati inequality and the additional uncertainty in the disturbance matrix results in a three-parameter Riccati inequality. The results presented in the paper are based on results by the author (see, for example, \textit{M. M. Kogan} [IEEE Trans. Autom. Control 43, No. 5, 670-674 (1998; Zbl 0920.93018)]) developed to solve the stabilization problem for continuous-time systems with norm-bounded time-varying uncertainty in state and input matrices without the matching condition.