ARGA loop pairing criteria for multivariable systems

An important aspect in the control of multivariable (MIMO) systems is the pairing of the input and output variables for decentralized control. This paper addresses the problem of pairing if performance of the system is degraded, or modified, as changes or abrupt variations occur in the interacting loops. Typical cases are stability of the process in the face of one or more loop failures - the integrity problem, or if one or more control loops are switched to `manual¿ by operators for maintenance and all other loops are maintained operating in stable conditions. Therefore a nonlinear, eventually time varying component is introduced inside each control loop to model such operations and the problem is solved in the framework of absolute stability theorems, mainly by using the Cook circle theorems. A global index, called Absolute Relative Gain Array (ARGA), is proposed for solving both integrity problems and robustness to parameter variations and nonlinearity effects. ARGA index has been tested on several examples from the literature.