XU Feng, PAN Qi, WANG Yilan, LUO Xionglin
The input-output variable pairing and the control-loop configuration are the first steps in designing a decentralized control configuration for multivariable processes. The relative gain array (RGA) method and improved variations have been widely used for control system designs, but they generally utilize time domain process information that create challenges when dealing with long lag times, integrators, and unstable systems. This work describes a relative frequency gain method as the ratio of the open-loop frequency gain to the close-loop frequency gain. An interaction measurement index is given through averaging the differences between the relative frequency gain and (1, j0) in the full frequency domain. The interaction measurement index array is used to create a new control-loop pairing criterion for the control structure selection. This method makes better use of the dynamic process characteristics than existing methods through frequency domain analysis. The method is simple and effective, with long lag times, integrators, and unstable systems.