Identification of closed-loop systems with low-order controllers (Q675291)

The paper deals with indirect identification of closed-loop systems. A bias correction method without noise modelling for the standard least-squares, introduced before by the author, is extended to the case when the order of the controller can be less than that of the open-loop plant -- in contrast to the previous development. This makes the method applicable to unbiased identification in relevant cases of closed-loop systems with low order controllers. The essence of the modification consists in adding a digital prefilter of the proper order at the input of the closed-loop system. The known controller parameters together with the prefilter parameters provide sufficient information for estimating the noise-induced bias. The resulting bias-correction scheme is not sensitive to common factors, the structure and level of the noise, requires moderate amount of computations, and yields consistent parameter estimates.