The iterative NCDE algorithm for ARMA system identification and spectral estimation

Abstract

The problem of identifying autoregressive moving average (ARMA) models with observational output data is addressed within this report. In the absence of actual input data, the ARMA identification problem is nonlinear in the parameters. The new general ARMA algorithm derived within, entitled NCDE, makes use of the Yule-Walker equations for input estimation and a least squares input-output ARMA algorithm for initial parameter estimation. The NCDE algorithm has been tested and results show that it is both effective and efficient for autoregressive (AR), moving average (MA) and ARMA system identification via the application of an ARMA model.