A model free dynamic criterion for online input-output pairing

In this paper, a new dynamic criterion for control configuration selection of stable and linear multivariable systems is presented. The proposed method is based on cross-covariance samples and can be computed from input-output data. According to the latter feature, the proposed approach is model free. It is also demonstrated that the proposed method regards both static and dynamic information of system simultaneously for measuring interaction. Furthermore, the proposed method can handle online input-output pairing in the presence of uncertainties. Finally, through examples the accuracy of proposed method is demonstrated.