Parametric identification of linear systems operating under feedback control

This paper considers the indirect identification of closed-loop systems with colored noise when low-order controllers exist in the loop. A new form of bias-eliminated least-squares (BELS) method is developed which is based on a simple and efficient technique of estimating the colored-noise-induced bias through a combined use of the known information about the controller and a proper number of zero parameters introduced. The main feature of the presented BELS based method is that no prefiltering of sampled data is needed and an unbiased estimate for the closed-loop parameters is given in a direct manner. The presented method is verified via application to indirect identification of an open-loop unstable plant.