Complex-coefficient ARMA parameter estimation based on positive frequency domain.

概要

論文の詳細を見る
Many control system design algorithms assume knowledge of the parameters of the signal process model. In practice rarely is there a priori knowledge of these parameters, and so there exists the need to identify a signal model first. In such a case an autoregressive moving-average (ARMA) model is frequently used.Durbin's method is an ARMA parameter estimation procedure, which provides good estimation with moderate computations. In this paper it is shown that the Durbin's method is not suitable for identifying the parameters of signals that possess lowpass and broadband power spectrum. Thus an alternative procedure is driven which could identify parameters of complex-coefficient ARMA model satisfying only the positive frequency characteristic of the unknown system. This procedure contains an efficient high-order AR model estimation algorithm, which is also proposed in this paper, and requires half the model order compared with the conventional Durbin's method. This algorithm can provide better estimation than the Durbin's one, upon finite precision arithmetic and with less computational burden. Computer simulations confirm those properties of the proposed procedure.
公益社団法人計測自動制御学会の論文