New identification method of nonlinear systems based on Hammerstein models

Hammerstein models are commonly used to present the pH process or the processes with characteristics such as exponent, dead-zone and switch. A new method for identification Hammerstein models is presented in this paper. Firstly, the nonlinear static part and the linear dynamic part are expressed by nonlinear basis functions and the Laguerre functions, respectively. The parameters of these two parts are then identified by least squares estimation, singular value decomposition and matrix dimension expansion. The convergence of the output error is also proved when white noises exist in the output signal. The proposed approach is based on weak assumptions on persistency of the excitation, so it is suitable for many applications. The operation is easy and the result is accurate. Finally, a simulation on a pH process is given to validate the conclusions.