LOW-ORDER CONTROLLER DESIGN BASED ON LMI USING PROJECTION METHOD

摘要

This paper presents a computational method to treat low-order suboptimal H{sub}∞ controller design problems described by linear matrix inequalities (LMIs). Using this method, a feasible pair (X, Y) with minimized trace can be found firstly andthen a projection onto reduced-rank parameter space is obtained in order to solving a reduced-order controller. The condition that a reduced-order controller stabilizes a closed-loop system is shown and an upper bound of the projection error is given. The robust performances of reduced-order control system is analyzed and a method of selecting nominal plant for a class of LTI system with interval parameters is proposed.