Data filtering‐based recursive identification for an exponential autoregressive moving average model by using the multi‐innovation theory

Abstract

This study employs the data filtering technique to investigate the recursive identification problems for a non-linear exponential autoregressive model with moving average noise, i.e. the ExpARMA model. Whitening the ExpARMA model by a linear filter, the original identification model is divided into a filtered identification model and a coloured noise model, then a filtering-based extended stochastic gradient algorithm is derived. In order to improve the parameter estimation accuracy, the multi-innovation identification theory is used to develop a filtering-based multi-innovation extended stochastic gradient algorithm for the ExpARMA model. A simulation example is given to demonstrate the superiority of the proposed filtering-based multi-innovation algorithm over the existing algorithms.