An efficient method for direct closed-loop identification

A novel type of least-squares (LS) based method in combination with the bias correction principle is proposed for direct identification of plants under feedback control. Its centerpiece is a more computationally efficient scheme for estimating the noise covariance vector that specifies the source of the noise-induced bias in the LS estimate. The attractive feature of the proposed method is that it can achieve the good estimation accuracy at a significantly reduced numerical cost. A numerical example is presented to demonstrate the effectiveness of the proposed method.