The Selection Of Disturbance Models In Performance Assessment Of Control Loops

Technique of online monitoring and diagnosis of control loops performance is a research hotspot of process industry automation at present. A real-time calculable performance index and corresponding assessment methods that do not disturb routine operation are required for online monitoring of control loops performance. Conventional control loops performance such as setpoint tracking performance and robustness performance is barely suitable for the design and tuning of controllers, but is not fit for online monitoring of control loops performance. The performance index and assessment technique proposed by Harris is a breakthrough research, which requires only closed-loop operational data with little prior knowledge——process time delay to evaluate the control loop performance. Its characteristic is quite suitable for the online monitoring of control loops performance. The performance index and assessment technique by Harris is the basis of performance assessment theory of control loops, and promoted the development of online monitoring and diagnosis technique of control loops performance deeply.Firstly, the performance-assessment theory and technique were investigated and summarized, which contained review of history, recent research directions as well as practical application cases. The objects of this research were introduced as well as the current research status in China.Secondly, the elementary theory of performance assessment was introduced. Minimum variance control law in univariate process was derived from the aspect of internal model control, and the implementation procedures of ARMA algorithm and FCOR algorithm are introduced. In simulation experiments, performance index was calculated by the above algorithms. Auto-correlation analysis of time series was also carried out to verify the assessment result.The influence of disturbance model on performance assessment result of SISO processes was studied. For classic first order/second order plus dead-time process and integral plus dead-time process, simulation experiments and performance assessment were carried out. The results were compared with the setpoint tracking performance as well as anti-disturbance performance. The minimum variance controller was derived under ARMA disturbance and IMA disturbance relatively to explain the experiment results logically, with the help of literature from MacGregor. Finally pointed out that sample time series must be filtered according to disturbance type, when applying performance assessment methods.The influence of disturbance model on performance assessment result of MIMO processes was studied. For a classic multivariate interactor plant——quadruple interactor tanks, simulation experiments were carried out, and methods of performance assessment for MIMO processes were applied to PID controller. It can be concluded that the influence of disturbance model on performance assessment in MIMO processes is similar to the one in SISO processes. The solution algorithm of unitary interactor matrix was simplified and implemented by Matlab script language. Proved the superiority of MPC in MIMO processes from the aspect of performance assessment.At last, the jobs were summarized and prospects of future research directions were introduced.