Artificial neural network-based algorithm for ARMA model order estimation

Abstract

This paper presents a new algorithm for the determination of the Autoregressive Moving Average (ARMA) model order based on Artificial Neural Network (ANN). The basic idea is to apply ANN to a special matrix constructed from the Minimum Eginevalue (MEV) criterion. The MEV criterion is based on a covariance matrix derived from the observed output data only. The input signal is unobservable. The proposed algorithm is based on training the MEV covariance matrix dataset using the back-propagation technique. Our goal is to develop a system based on ANN; hence, the model order can be selected automatically without the need of prior knowledge about the model or any human intervention. Examples are given to illustrate the significant improvement results. © 2010 Springer-Verlag Berlin Heidelberg.