Self-Tuning Control: Algorithms and Applications

摘要

The long term integrity of a self tuning controller is considered. Minimum-variance and pole-placement control approaches are developed from a CARIMA model, rather than from CARMA representation of a process. This is shown to introduce integral action into the control laws, and produces estimation schemes with a zero mean structure. Data conditioning and control signal saturation are considered, and jacketting routines are proposed. Rules for commissioning a self tuning controller are introduced. Two variable forgetting factors are compared in simulation and on a full scale industrial process, and are shown to improve the reliability of a self tuning controller. Application studies which involve cascaded self tuning controllers are presented.