Robust model predictive control of stable linear systems. (English) Zbl 0889.93025

Badgwell, Thomas A.

Int. J. Control 68, No. 4, 797-818 (1997).

This paper presents a new robustly stabilizing model predictive control (RMPC) algorithm. Model uncertainty is parametrized by a list of possible plants. The basic approach described in the paper can be used to robustly stabilize a wide class of stable nonlinear processes with many possible controller cost functions. For the special case of regulating stable, linear plants with hard input and soft state constraints and with a quadratic objective, the algorithm reduces to a quadratic program with an additional constraint that is quadratic in the input vector. The significant advantage of the RMPC algorithm is that no new tuning parameters are necessary to achieve robustness. The results of the numerical simulation for two examples are presented.

Reviewer: I.Randvee (Tallinn)

MSC:

93B51	Design techniques (robust design, computer-aided design, etc.)
93D21	Adaptive or robust stabilization

Keywords:

robust stabilization; model predictive control; nonlinear processes