Gain-scheduled H2 non-minimal order filtering design for linear parameter-varying discrete-time systems

Abstract

This paper addresses the problem of gain-scheduled filter design with guaranteed H2 performance for linear parameter-varying discrete-time systems with arbitrarily fast time-varying parameters. The filters to be designed have order greater than the order of the system being filtered. The design procedure is formulated in terms of Linear Matrix Inequalities. As is well known, gain-scheduled filters can provide superior performance for linear parameter-varying systems when compared with robust filters. The proposed non-minimal order H2 gain-scheduled filtering strategy can further improve the results by enlarging the structure of the filter. Examples illustrate the results, showing that larger order filters can have improved performance, providing lower guaranteed H2 costs when compared with other approaches based on full order filtering from the literature.