Process model quality monitoring of model predictive controller

Abstract

Model quality is one of the decisive factors that affect the performance of model-based controllers. However, the existing methods were devoted to the monitoring of the overall model quality, and didn't distinguish mismatch of the process model and the disturbance model. Aiming at this problem, this paper presents a new method for monitoring the quality of process model based on input and output data. The proposed method estimates the disturbance innovations using routine operation data according to the feedback invariance principle of disturbance innovation in closed-loop control system, where the setpoints are time variant, and takes the ratio of a quadratic form of model residual and extended form of estimated disturbance innovations as an index to assess the process model quality. Combining the proposed index and the existing overall model quality index, the process model mismatch and the disturbance model mismatch can be separated successfully. The effectiveness of the proposed method is verified by the simulation study of Wood-Berry binary distillation column.