Performance Monitoring of Economic Model Predictive Control Systems

详细信息    查看全文

• 作者：Matthew Ellis ; Panagiotis D. Christofides
• 刊名：Industrial & Engineering Chemistry Research
• 出版年：2014
• 出版时间：October 8, 2014
• 年：2014
• 卷：53
• 期：40
• 页码：15406-15413
• 全文大小：451K
• ISSN：1520-5045

文摘

A framework for performance monitoring of economic model predictive control (EMPC) systems is presented which includes the computation of an acceptable operating region, which is a well-defined region in state-space, for EMPC systems to operate a process in a time-varying fashion to optimize process economics while meeting input constraints and stabilizability requirements. To capture the interplay between sources of common cause variance caused by various sources like sensor noise, imperfect actuator operation, and model inaccuracy, a residual variable taken to be the difference of actual real-time economic cost and the predicted (expected) economic cost is defined. Utilizing exponentially weighted moving average (EWMA) and historical closed-loop process data, an upper control limit and a lower control limit are established which defines normal operation (i.e., operation with common cause variation). The limits are utilized to monitor the performance of EMPC by comparing real-time process operation data under EMPC and the corresponding regions of acceptable EMPC operation computed in the normal operation dynamic data generation step. The proposed monitoring framework is demonstrated and evaluated using a chemical process example.