Indirect closed-loop identification by optimal instrumental variable method (Q1388125)

In system identification, the instrumental variable (IV) method provides a promising way to achieve consistent parameter estimation for systems corrupted by colored noises. The advantage of the IV method is that its estimation results do not depend on the accuracy of the noise models. Recently, a new bias-compensating least-squares (BCLS) method was proposed for the identification of a closed-loop system with high-order controller. This paper studies the connection between the BCLS method and the instrumental variable (IV) family. It is shown that the BCLS method is a kind of weighted instrumental variable (WIV) method whose results do not depend upon the order of the controller. It is also shown that the restriction on the order of the controller used in the BCLS method can be relaxed in the presented optimal IV method.