Abstract
This paper is concerned with the problem of identifying parameters in single input-single output linear discrete-time systems where the output is corrupted by additive noise. A recursive identifier is developed by using the techniques of a linear approximation filter and by using the positive real transfer function introduced by Landau. The usage of the transfer function assures the convergence of the estimates of parameters and also improves the rate of their convergence. It is shown that a recursive maximum likelihood identifier arrives at the same form with the proposed identifier if its estimation model is corrected slightly. Numerical results obtained from digital simulation in a forth-order system are presented to illustrate the utility of the proposed identifier.
