Stable Generalized Predictive Control: An Algorithm with Guaranteed Stability

摘要

An idea that removes the stability problem but retains and further simplifies themain features of Generalized Predictive Control (GPC) is proposed. In particular, a characterization of the class of all stabilizing feedback controllers is used as a first step in order to achieve internal stability. The output prediction and optimization steps used in GPC are then performed but the degrees of freedom are defined to be future values of the reference signal rather than values of future control increments. The stabilizing feedback configuration and the algorithms are described. Treatment of robustness is presented. Results are illustrated by means of three numerical results.