Multistage Hammerstein-Wiener System Identification with the Help of Binary Excitation

In the paper, we consider a problem of identification of HammersteinWiener (N-L-N) system. The proposed strategy consists of 4 steps. Firstly, we identify the finite impulse response of the linear block in the presence of random input and random noise using both parametric and non-parametric identification tools (least squares and kernel estimate). Secondly, special binary sequences are generated to recover second nonlinear characteristic. Next, the interactive signal between dynamic linear and second nonlinear block is estimated. Finally, we recover first nonlinear characteristic using standard algorithm for Hammerstein system identification (least squares and singular value decomposition).