A sufficient asymptotic stability condition in generalised model predictive control to avoid input saturation

Abstract

The goal of this contribution is presenting a Theorem which states the asymptotic stability of a feedback controlled system with a Linear Generalized Model Predictive Control (LGMPC). Concerning the asymptotic stability, a sufficient and constructive condition on the weight matrices of the cost function used in the optimization problem in LGMPC for one step prediction horizon is demonstrated. The condition consists of a lower bound for one of these matrices. The obtained condition is explained and discussed by means of some physical considerations. The second part of this contribution is devoted to the saturation case and proves a sufficient condition for obtaining asymptotic stability and saturation avoidance.