Closed-loop Identification and Periodic Disturbance Estimation Based on Generalized Minimum Variance Evaluation

This study addresses a closed-loop identification problem using routine operation data that include influences of periodic and stochastic disturbance. By using the periodicity, the proposed approach can estimate both model parameters and periodic disturbances. The paper also gives a new data-driven criterion and shows that its optimization uniquely obtains the actual model parameters and periodic disturbances. A simulation example demonstrates the validity of the proposed method.