Robust predictive control of systems with parametric uncertainties

This paper presents a regulation scheme capable of achieving the nominal tracking performance provided by the constrained receding horizon predictive control (CRHPC) together with quadratic stability in the face of parametric model uncertainties. A two-step procedure is followed. First, uncertainty is neglected and a CRHPC regulator is synthesised for the nominal plant. Then, a "robustifying" controller is designed, using the standard H_∞ approach, for the augmented system formed by the true system, the first-step predictive controller and an internal model of the nominal plant.