Parameter estimation for ARMAX systems using bias compensation methods

For ARMAX systems, this paper derives a bias compensation recursive least squares (BCRLS) identification algorithm by means of the prefilter ieda and the bias compensation principle. The proposed algorithm realizes the recursive computation of the bias compensation methods and can be on-line implemented. The BCRLS algorithm can give the unbiased estimation of the system model parameters in the presence of colored noises, irrespective of the noise model. Finally, the advantages of the proposed BCRLS algorithm over the non-recursive bias compensation least squares (BCLS) algorithm are shown by simulation test.