Asymptotic statistical accuracy analysis of virtual reference feedback tuning control.

Virtual reference feedback tuning control is a data-driven control strategy. No model identification of the plant is needed in this control method. Now virtual reference feedback tuning control is a widely used procedure for controller tuning process. Because the asymptotic covariance matrix expression is an important factor in the whole system identification theory. This paper derived the error about the unknown parameter estimation through Taylor series expression. Then it established the corresponding covariance matrix of the parameter estimation error. It obtained the two diagonal sub-matrices in the covariance matrix using some matrix operations. These two diagonal sub-matrices were the asymptotic covariance matrix expression of the two unknown parameter estimation vectors in the closed-loop system. Based on this asymptotic covariance matrix, it obtained an optimal filter by solving an optimization problem which included some trace operation. Finally, the efficiency and possibility of the proposed strategy can be confirmed by the simulation example results. [ABSTRACT FROM AUTHOR]