Maximizing the stability radius: an LMI approach

Given a stabilizable linear system Ex/spl dot/ = Ax + Bu with sE - A regular, we analyze the stability robustness of the closed-loop system (E + BK) = (A + BF)x + v, obtained by proportional and derivative (PD) state feedback u = Fx Kx/spl dot/ + v. Our goal is to maximize the stability radius of the closed-loop system matrix s(E + BK) - (A + BF) over all stabilizing PD state feedback control laws. This problem turns out to be equivalent to a particular H/sup /spl infin//control problem for a generalized state-space system and reduces to a system of matrix inequalities. Under certain conditions the problem actually reduces to an LMI system. We also show how to apply these ideas to higher order dynamical systems.

[1]  András Varga,et al.  Minimal Degree Coprime Factorization of Rational Matrices , 1999, SIAM J. Matrix Anal. Appl..

[2]  Vincent Vermaut,et al.  On stability radii of generalized eigenvalue problems , 1997, 1997 European Control Conference (ECC).

[3]  Chengwu Yang,et al.  H∞ state feedback control for discrete singular systems , 2000, IEEE Trans. Autom. Control..

[4]  P. Van Dooren,et al.  Real and complex stability radii of polynomial matrices , 2002 .

[5]  Pierre Apkarian,et al.  Robust pole placement in LMI regions , 1999, IEEE Trans. Autom. Control..