Gaussian Process Model Based Performance Assessment of Control Loop

Abstract

Practical constraints on control systems such as valve constraints imply that performance of minimum variance control is most likely not achievable. Instead, the objective may be expressed as minimizing a linear quadratic function of input and output variances instead. The evaluated performance index is dependent on the model and in terms of accuracy of the model it is only as accurate as the quality of the data available. This work proposed a Gaussian process model indicator to provide a trust region on the performance index and to provide a framework of judging what improvement is feasible. The model can be updated adaptively and in turn the performance. Based on the indication of the model quality the reliability of the evaluated controller performance index can be obtained.