Two-stage Gradient-based Recursive Estimation for Nonlinear Models by Using the Data Filtering

This paper considers the parameter estimation problem of a two-input single-output Hammerstein finite impulse response system with autoregressive moving average noise. Applying the data filtering technique, the input-output data is filtered and the original system with autoregressive moving average noise is changed into the system with moving average noise. Then, based on the key term separation technique, the filtered system is decomposed into two subsystems: one subsystem contains the unknown parameters in the nonlinear block, the other contains the unknown parameters in the linear dynamic block and the noise model. A filtering based multi-innovation stochastic gradient algorithm is presented for Hammerstein finite impulse response systems. The simulation results confirm that the proposed algorithm is effective in estimating the parameters of two-input single-output Hammerstein finite impulse response systems.