A New Framework for Controller Performance Assessment

Abstract

A new framework of controller performance assessment is presented. The control performance index -relative variance index is recalculated more simply by adopting the normal operating data plus knowledge of the plant model. Then, the potential improvement to control performance by retuning the controller can be detected according to a comparison of the current control performance with that of the optimally tuned controller, which is based on the regeneration of normal operating data accuracy without estimating the disturbance model, instead, estimating disturbance directly. This helps define the best controller for specific systems, in order to make immune from any exceptional disturbance. The benefits of this new control performance assessment framework are demonstrated on a univariate system simulation.