A Modification on Stochastic Approximation Algorithm with Expanding Truncations For Estimation of NARX Systems

In this paper, nonparametric identification for nonlinear autoregressive systems with exogenous inputs (NARX) is considered. In order to provide the identification of the nonlinear system, a recursive stochastic approximation algorithm with adaptive thresholds with fuzzy observations (modified SAAWET) is used. The effectiveness of the proposed method is proven through implementation of algorithm on two systems. It will be figured out that the proposed algorithm will result in better estimation and robustness in comparison to conventional SAAWET.