Predictive Controllers with Guaranteed Stability and Mean-Level Controllers for Unstable Plant

Model-based predictive controllers do not have guaranteed stability in general; however, in recent years many predictive control strategies have been adapted to give guaranteed stability. Three of the contributions to this study have been shown to have a common end-point constraint and hence give identical control laws. The end-point constraint, however, was normally satisfied indirectly. Here it will be shown how explicit equations for the system predictions can be derived that satisfy that end-point constraint. As the equations are explicit they are computationally efficient. In the original generalised predictive control (GPC) algorithm, guaranteed stability was available for stable plants using an infinite output horizon and a finite input horizon; this was called mean-level control. In this paper it is shown that using a less stringent end-point constraint to that common in predictive control strategies with guaranteed stability, a mean level result exists for unstable plant and is of low computational burden. Again explicit prediction equations are derived.