Nominal and Robust Stability in Receding-Horizon Predictive Control

D. Megías and J. Serrano

Abstract. This paper is focused on the robustness of receding-horizon predictive control with end-point equality constraints. The use of these constraints ensures nominal stability in this class of controllers. However, since there is always a difference between the internal model and the real plant (system uncertainty), closed-loop instability may arise. A method of increasing robustness based on softening the constraints (-design) is suggested. Its properties are studied and compared with other ways of enhancing robustness. Simulation results show that the -design prevents instability while good performance is preserved.

Key words. predictive control, uncertainty, robustness, end-point constraints.

Registration: PIRDI-6/98, July 1998.

Retrieve PostScript document ( 307633 bytes)
Copyright © 1998 by Computer Science Department. UAB