Iterative Identification of Neuro-Fuzzy-Based Hammerstein Model with Global Convergence

In this paper, a neuro-fuzzy-based model is used to describe the nonlinearity of the Hammerstein process without any prior process knowledge, thus avoiding the inevitable restrictions on static nonlinear function encountered by using the polynominal approach. In doing so, a clustering algorithm is presented in order to identify the centers and widths of the neuro-fuzzy-based Hammerstein model, and an updating algorithm guaranteeing the global convergence of the weights of the model is developed based on the Lyapunov approach. As a result, the proposed method can avoid the problems of initialization and convergence of the model parameters, which are usually resorted to a trial and error procedure in the existing iterative algorithms used for the identification of Hammerstein model. Examples are used to illustrate the performance and applicability of the proposed neuro-fuzzy-based Hammerstein model.