Performance Assessment Of Control Loop With Time-variant Disturbance Dynamics

At the beginning of the operation in industrial process, the control system often can run in good condition, but as time goes on, the performance of the control system is often unable to maintain the original optimal state. The control loop which has defects may results in many problems. More and more experts and scholars have realized the importance of performance assessment, and have obtained certain research results in this field. But there is still some problem in this field, such as multiple model predictive control has been more and more used in actual industry, but now, the research of how to assess the performance this kind of control system has not begun; the existing methods can only tell the system is good or not, but can’t tell what cause the system performance degradation; most of the performance assessment research focus on time invariant system, but the research on how to assessment the performance of time-varying system is relatively less.In the view of the problem left over, in this paper, I do some relevant research based on the predecessors’ research. The main research work and innovation are as follows:1. There is a kind of system that the parameter changes abruptly, so a generalized predictive control algorithm based on multiple models switching is presented to replace the traditional single model generalized predictive control algorithm, based on the Filtering and Correlation Analysis Algorithm to assess the performance. The simulation results show that multi-model predictive control system has better transient performance and steady-state performance, they also show the effectiveness of the performance assessment method based on FCOR algorithm.2. In view of the multivariable predictive control system with constrains, this paper present based on the evidence network method to evaluate the performance and diagnosis. Evidence network method can quantify uncertainties information in the system caused by incomplete information and the information is not accurate, more accurate assessment the performance of system, and diagnosis of the cause of system performance degradation. Industrial simulation shows that the method better than the traditional method based on Bayesian networks.3. In view of the time-varying disturbance control system, this paper present a method based on multi-model mixing minimum variance to assessment the performance. This method fully considering the feature of each disturbance, enhance the accuracy and effectiveness of the evaluation results, avoid the intermittent switching and larger transient error caused by the multi-model switching method. Numerical simulation and industrial simulation show the feasibility and superiority of this method.