Design of robust controller for linear systems with Markovian jumping parameters (Q1289365)

The problem considered is the robustness for a class of uncertain linear systems with Markovian jumping parameters: \[ \begin{aligned} \dot x(t) & = (A(r(t))+\Delta A(r(t),t)) x(t)+(B(r(t))+ \Delta B(r(t),t)) u(t),\\ y(t) & = (C(r(t))+\Delta C(r(t),t)) x(t)+ (D(r(t))+\Delta D(r(t),t)) u(t),\\ x(0) & = X_0,\end{aligned} \] where \(\{r(t), t\in [0,T]\}\) is a homogeneous and right continuous Markov process, taking values on the finite set \({\mathcal B}= \{1,\dots, N\}\), and the matrices \(\Delta A(\cdot)\), \(\Delta B(\cdot)\), \(\Delta C(\cdot)\) and \(\Delta D(\cdot)\) represent uncertainties, and appropriate dimensions for all variables and matrices are understood here [cf. \textit{Y. Ji} and \textit{H. Chizeck}, IEEE Trans. Autom. Control 35, No. 7, 777-788 (1990; Zbl 0714.93060)]. Under the assumptions of boundedness of the uncertainties, a sufficient condition for robust stochastic stability of this class of systems is established. It involves \(N\)-coupled algebraic Riccati equations [cf. \textit{I. R. Petersen}, Syst. Control Lett. 10, No. 5, 341-348 (1988; Zbl 0663.93022)]. An iterative algorithm, together with a numerical example to determine the parameters of the robust controller are given in detail.