Robust performance of a class of uncertain systems (Q4946755)

The author examines the problems of robust performance and feedback-control synthesis for a class of linear systems with time-dependent parametric uncertainties, described as \[ \dot x(t)=A(\varDelta)x(t)+B(\varDelta)w(t),\quad x(0)=0,\quad z(t)=C(\varDelta)x(t)+D(\varDelta)w(t), \] where, at time \(t\in\mathbb R_+\), \(x(t)\in\mathbb R^n\) is the state vector; \(w(t)\in\mathbb R^q\) is the disturbance input vector; \(z(t)\in\mathbb R^q\) is the controlled output; and \(A(\varDelta)\), \(B(\varDelta)\), \(C(\varDelta),\) and \(D(\varDelta)\) are continuous bounded matrices whose entries are functions of \(\varDelta(t)\); \(\varDelta(t)\in\varXi\) is a (possibly) time-dependent uncertain matrix assumed to be Lebesgue measurable. Using the recently developed criterion of strongly robust \(H_\infty\) performance, some earlier results on stability and performance are recovered and then extended to some new and easily computable linear matrix inequalities. These inequalities include the algebraic Riccati inequality, the Lyapunov inequality, and a linear matrix inequality. Synthesis of robust feedback controllers is carried out for several system models of interest.