Interaction and Controller Selection in Decentraliced Control of TITO Systems: IMC Based PI/PID Approach

Performance of multivariable control systems may be highly degraded because of the existence of interaction terms. It is therefore important to have measures that help in the controller selection procedure. Special attention has to be paid to the potential instability caused by interaction effects generated by the control action on the other loop. This paper provides an analysis by means of an interaction measure defined in terms of the expected interaction to be generated into the control systems in terms of the desired specification for the local control loops, and shows how this can help to select the appropriate controller. Detailed presentation concentrates on the two-input two-output case and the controller design using the IMC method.