Adaptive-Robust Control of Dynamic Systems Using Generalized Predictive Methods

This paper considers an approach to adaptive-robust control of dynamic systems using generalized anticipation, where the object is described by a locally linearized model. Known self-tuning algorithms show insufficient stability to inaccurate selection of delay or model order. A generalized predictive control approach is suggested, with simulation outcomes showing superiority over traditional methods like generalized minimum variance control and pole assignment.This sliding horizon algorithm is based on predicting future system output signals several steps ahead, based on assumptions about subsequent control actions. One effective assumption is the presence of a “control horizon,” beyond which control signal increments are assumed to be zero. This condition contributes to both increased stability and simplified computational procedures. The selection of forecast and control horizon values allows for various useful algorithm variants to be obtained as special cases. The results obtained are highly effective when using adaptive controllers to control objects with varying delays, orders, and parameters.