Control performance assessment of multivariable system based on multi-time-variant-disturbances mixing GMV method

The disturbance in chemical process is complex and has the multiple characteristics, and the control performance assessment of multivariable system with multiple disturbances is one of the hot topics. In this paper, the control performance assessment method of multivariable systems, based on multi-time-variant-disturbances mixing generalized minimum variance (MMGMV), is proposed. Firstly, the generalized minimum variance control is introduced into the multivariable system performance assessment, and the weight matrix is designed according to the time-varying control object. Then, the multivariable MMGMV controller is designed combining with the idea of multi-model weights mixing for all multi-time-varying disturbances. Next, the output variance of each controlled variable is obtained using MMGMV controller. The average variance of controlled variable in the MMGMV controller acts as the criterion of performance assessment, and combining with the output variance of actual controller for the controller performance assessment. Compared with the minimum variance benchmark, the developed method is more reasonable and practical for the control performance assessment of multivariable systems. The developed approach is demonstrated by a numerical simulation and a heavy oil fractionation of process control system.