Multi-signal source identification of ELM hammerstein model with colored noise
Zhenzhen Han, Bin Cheng, Yunli Wang, Yunxia Shao · 2018
Hammerstein model is used in practical industrial process widely. A novel ELM-Hammerstein model with colored noise is proposed, where the extreme learning machine (ELM) is used to describe static nonlinear part and the dynamic linear part is described by CARAR model. The purpose is to identify the parameters of ELM-Hammerstein model. But, intermediate signal can not measure directly in the process of identification. So special signal is employed to separate the static nonlinear part and the dynamic linear part of the Hammerstein model. Further, recursive extended least squares (RELS) algorithm is applied to compute the unknown parameters of linear part. As a result, the proposed method can describe the nonlinear system with colored noised with high accuracy. Simulation example demonstrates its effectiveness.