Identification of MISO Hammerstein Nonlinear Model with Moving Average Noise Based on Hybrid Signal

The hybrid signal is used to identify the multi-input single-output (MISO) Hammerstein model. The hybrid signal consists of Gaussian signal and random signal, and the identification process is divided into two stages, namely, the stage of dynamic linear part and the stage of static nonlinear part. Firstly, the correlation analysis method is used to identify the linear part parameters. Then, for the parameters of the nonlinear part and the output noise model, an extended stochastic gradient algorithm with forgetting factor (FF-ESG) is adopted to deal with the issue that the convergence of stochastic gradient algorithm is slow. Theoretical analysis and experiments show that the presented method can identify the MISO Hammerstein model with moving average noise and obtain good identification accuracy.