Robust Output Regulation using Reduced-Order Controllers

Abstract

In previous articles, robust regulator problem has been studied. It has been shown that an optimal robust controller (a solution to regulator problem) can be approached by using H ∞ -optimization and jω -axis shifting techniques. A problem with this approach is that the controller might have a high MacMillan degree which is n + 2pl – q , where n is the degree of the nominal plant P 0 (s) є (R(s)) p×m , l is the number of the unstable modes contained in the external reference input and the disturbance and q is one in general. In (Wu and Mansour 1990b) it was shown that there are always stable cancellations between the two parts of the controller: the servo-and the stabilizing compensators and the value n + pl – q is a degree bound for the controller. In this paper, the problem of using reduced-order controllers to achieve robust output regulation will be discussed.