Closed-loop system identification by residual whitening

The accuracy of the system model during the identification process for dynamic systems under closed-loop operation is studied. The simulation results show that most of the model error comes from the approximated system description by using a finite difference model called the autogressive with exogeneous input model. Then the residual whitening method is proposed to improve the model accuracy. An iterative procedure for minimizing and whitening the residual of the autogressive moving average with exogeneous input (ARMAX) model is provided. Numerical simulations of a highly unstable, large-gap magnetic suspension system are presented to validate the proposed residual whitening method.