Robust model predictive control of constrained systems

Robust analysis and synthesis of model predictive control (MPC) are of significant theoretical and practical importance. One contribution of this thesis is the analysis of the inherent robustness properties of quasi-infinite horizon MPC. For constrained nonlinear systems, it is one of the most important MPC schemes with guaranteed nominal stability. We prove that quasi-infinite horizon MPC for systems with input constraints has some inherent robustness. The analysis proposed in this thesis does not depend on the continuity of the optimal cost functional, as was the case in previous studies in the literature. The other contribution of this thesis is synthesis for robust MPC. Since it is difficult to propose a general MPC scheme for systems with respect to all types of disturbances or model/plant mismatches, we consider three special kinds of constrained systems: nonlinear systems with respect to persistent disturbances, linear systems with structured feedback uncertainty, and linear parameter varying systems. Both recursive feasibility and robust stability are guaranteed for the proposed schemes.