On-line solution of system constraints in generalized predictive control design

Abstract

A specific efficient solution of constrained system outputs is presented. The solution is based on on-line weighting of individual system outputs or output components representing in fact a specific on-line tuning/applification of the components of weighting control parameters. Thus, the predictive control design retains flexibility without any increase of time-computational demands. The proposed solution represents only soft solution of the constraints, i.e. a compliance with the constraints is not fully guaranteed, but the result is satisfactory and not so far from results of hard-constraint solution computed e.g. by a quadratic programming. The solution predetermines the area of the use to the systems with fast dynamics requiring very high sampling rates as electric drives and the like. The solution is presented in the paper as a generalized solution for common use. It is demonstrated by simple simulative examples and one illustration from a real control of 3-phase electric drive.