Abstract：By designing the input signals of the industrial large-scale system, the linear part of the Hammerstein model can only be reflected by system output, and the nonlinear static effect can be effectively removed from the system. The linear part of the system can be identified by using the decentralized identification method. Based on the obtained linear mode and using the measured system outputs, the intermediate inputs can be reconstructed. So the nonlinear part parameters can be estimated. The dynamical separation method is proposed for the industrial large-scale system Hammerstein model identification.The simulation results show its effectiveness of the method.