A State-Space Approach to Robustness Analysis and Synthesis for Nonlinear Uncertain Systems

Abstract

A state-space characterization of stability and performance robustness analysis and synthesis with some computationally attractive properties for nonlinear uncertain systems is proposed. The robust stability and robust performances for a class of nonlinear systems subject to bounded structured uncertainties are characterized in terms of various types of nonlinear matrix inequalities (NLMIs), which are natural generalizations of the linear matrix inequalities (LMIs) that appear in linear robustness analysis. As in the linear case, scalings or multipliers are used to find storage functions that give sufficient conditions for robust performances; these are also necessary under certain assumptions about smoothness of the storage functions and structure of the uncertainty. The resulting NLMIs yield convex optimization problems. Unlike the linear case, these convex problems are not finite dimensional, so their computational benefits are far less immediate. Sufficient conditions for the solvability of robust synthesis problems are developed in terms of NLMIs as well. Some aspects of the computational issues are also discussed.