Off-line output feedback robust MPC with general polyhedral and ellipsoidal true state bound

Abstract

In this paper, a novel off-line output feedback model predictive control strategy for linear parameter varying model is presented. This approach avoids the on-line optimization of the control law parameters, and simplifies the online computation mainly to refreshing the true state bounds (TSBs). In two previous works, the off-line output feedback model predictive control with ellipsoidal real-time TSB, and the on-line one with general real-time TSBs (both ellipsoidal and polyhedral), have been proposed. In this paper, the merits of both previous works are merged, so that the control performance can be greatly improved, with moderate increase of computational burden. Appropriate modifications in both on-line and off-line stages of the off-line MPC are taken, so that the closed-loop system is shown to be convergent. The effectiveness of the proposed approach is verified through numerical examples.