Performance Assessment Of MIMO Processes Based On Minimum Variance Benchmark

Due to the change of process operating conditions, time variant external disturbances and problems associated with instrumentation and equipment, industrial process controllers cannot achieve the design performance and even seriously deteriorate after running for a period of time. Therefore, it is a concerned problem for local technicians and control engineers that how to obtain the performance information of the control systems from the routine operating data so that the performance problem can be recognized and diagnosed early. The Harris index proposed for feedback controller performance assessment greatly activates the study in this field. However, for multivariable control system, the process time delay is a interactor matrix, that means the Harris index cannot be achieved by analyzing the closed-loop system output data directly. In order to solve this problem, Huang have proposed the FCOR (Filtering and CORrelation analysis) algorithm, firstly the unitary interactor matrix should be calculated by using the Markov parameters matrices of the process transfer function matrix, secondly the interactor-filtered output variable can be obtained from the unitary interactor matrix and the closed-loop system output data, finally, the Harris index for multivariable control system will be obtained by analyzing the interactor-filtered output variable.The key problem using Harris index to assessing the performance of control system is how to obtain a performance benchmark of minimum variance control for control system, the FCOR algorithm provide a simple and effective method to solving this problem, but the FCOR algorithm was proposed based on a condition, that the disturbance is a stationary processes. In fact, for MIMO process in actual industrial, the disturbance is usually nonstationary, and each noise might be different. In this paper, an approach for assessing the performance of multi-input multi-output processes under a special disturbance, which is mixed by stationary signal sources and periodic signal sources. As we all know, for multi-input multi-output linear processes, the system output is nonstationary when the disturbance is nonstationary. Thus, we will handle the system output data by using the difference method, and then the interactor-filtered output variable will be obtained and used to estimate the performance benchmark of minimum variance control for this system with FCOR algorithm. The simulation results prove that, for a MIMO process under a nonstationary disturbance mixed by stationary signal sources and periodic signal sources of various parameters, if the system output data is handled with the difference method, then the performance benchmark of minimum variance control by the FCOR algorithm from the new time series is consistent and accurate. So the approach of difference method for nonstationary system output is effective.This paper firstly gives the background, meanings, research status and relevant knowledge of performance assessment of control systems. Then we give some preliminaries for a performance assessment approach using a performance benchmark of minimum variance control, this approach is a main analytical method of this text, it runs through this text all the time. In the following the FCOR algorithm for multi-input multi-output process is introduced primarily, including the FCOR algorithm flow chart, estimation of the unitary interactor matrix, the interactor-filtered output variable and the modeling of multivariable linear control system. Finally, we analysis the performance assessment of MIMO processes under a nonstationary disturbance with MATLAB.