The design of servomechanism controllers for uncertain plants

Abstract The design of multivariable servomechanism controllers is considered here for plants whose parameters are uncertain or perturbed within a cone-bounding criterion. The solution comes from a controller design problem as an LQ optimal control problem in which the gain matrices result in a quadratic performance upper bound (similar to that described by Gilman and Rhodes (1976() which depends on the uncertainty. Under conditions on the nominal plant and on the uncertainty, a resulting linear time-invariant controller is guaranteed to achieve the objectives of robust tracking and regulation with closed-loop stability.