GMV and restricted-structure GMV controller performance assessment multivariable case

The application of control loop performance assessment and benchmarking techniques to multivariable industrial process control loops is considered. The results for assessing the performance of multiple-input, multiple-output (MIMO) control loops against the generalized minimum variance (GMV) benchmark, using routine operating data and the knowledge of the interactor matrix are presented. Then, assuming knowledge of the system model, the optimal controller is restricted to be of a low-order classical structure so that a more realistic benchmark is obtained. The technique may also be used to determine the best structure to use for a multivariable controller. This paper presents an extension of the existing results to the cases of multivariable data-driven GMV benchmarking and multivariable model-based RS-GMV benchmarking.