Research On Minimum Variance Performance Assessment Method Of Control System

As contemporary industrial processes continue to grow,more and more control loops are included in the plant,and sometimes a plant contains hundreds or even thousands of control loops.The number of these control loops is large,and after a period of operation,the performance of the control system is degraded due to changes in operating conditions during operation,non-linearity of the process,and failure of the machine equipment,resulting in a decrease in the quality of the produced product.Therefore,research on the performance assessment of control systems in recent years has been increasingly valued by scholars at home and abroad.Based on the background of the performance assessment and historical development,this article improves on the research results of the predecessors and conducts in-depth research on the minimum variance performance assessment algorithm of the control system.The focus of the performance assessment of control system is to select the appropriate algorithm to calculate the reference value.Therefore,it is necessary to obtain the system model and the delay information.Firstly,this article introduces time series analysis,which has a wide range of applications in the performance assessment,monitoring and fault diagnosis of controlled objects.It mainly includes:data stability test and preprocessing,three stationary time series models:autoregressive(AR)model,moving average(MA)model and autoregressive sliding average(ARMA)model,model order and two parameter estimation methods.Then,for the typical single-variable system with a single disturbance,aiming at the shortcomings of the traditional linear regression algorithm,the reference forgetting factor and the recursive least squares method are proposed to improve the performance index;On this basis,it is extended to the system with multiple disturbances,In the control system,the minimum variance performance index based on feedback invariant is proposed.For the system that sets the input to change with time,the method of least square fitting is proposed to improve the performance index.Finally,from the univariate expansion to the multivariable system,the filter and correlation analysis(FCOR)algorithm is theoretically derived to obtain the minimum variance performance index of the multivariate system.At the same time,to simplify the calculation,the diagonal matrix is introduced into the calculation of the reference value.A new performance assessment method based on the diagonal correlation matrix is proposed.