A simple design method of H∞ reduced-order filters for stochastic systems

Abstract

This paper is concerned with reduced-order /sub /spl infin// filtering of stochastic systems. Based on linear matrix inequality (LMI) technique, a new design method is proposed for the reduced-order filtering of stochastic linear systems. The method is derived from decomposing the key matrix in LMIs which determines the order of designed filters. Different from the existing methods, the proposed method first minimizes the upper bound of the key matrix and then eliminates its near-zero eigenvalues, which results in a simpler, more direct and reliable design procedure. The method is applicable to both continuous and discrete time stochastic systems. Its effectiveness is illustrated by an example.