A frequency-domain iterative identification algorithm using general orthogonal basis functions (Q5936360)

A two-step frequency-domain identification algorithm is proposed. For this purpose, the identified model is parametrized in terms of general orthonormal basis functions. An up to date scheme is presented where the poles of the basis functions are adjusted iteratively and the convergence properties with respect to unknown (bounded) noise are studied. The algorithm is applied to experimental frequency response data obtained from a compact disk player. Under some restrictions, the algorithm decreases the bias error to the order of the noise amplitude. This analysis requires the knowledge of the true order of the system at hand.