On-line control performance monitoring of MIMO processes

Abstract

The main purpose of this paper is to introduce a new method of control loop performance analysis of the closed-loop MIMO process. This new approach is based on filtering and subsequent correlation (FCOR) analysis of the output and the filtered data. The steps required in the implementation of the FCOR algorithm are: (i) estimation of the time delay or the interactor matrix of the MIMO system and (ii) evaluation of control performance with MIMO minimum variance control as a benchmark standard. The proposed algorithm is evaluated by application to a simulated and an industrial process.