HAMMERSTEIN MODEL IDENTIFICATION BY MULTILAYER FEEDFORWARD NEURAL NETWORKS

A new method for the identification of the nonlinear Hammer stein mode l, consisting of a static linearity in cascade with a linear dynamic p art, is introduced. The static nonlinearity is modelled by a multilaye r feed forward neural network (MFNN) and the linear part is modelled b y an autoregressive moving average (ARMA) model. A recursive algorithm is developed to update the weights of the MFNN and the parameters of the ARMA. The new method makes use of the well-known nonlinear mapping ability of MFNN and avoids the restrictive assumptions of the previou s identification methods. Two numerical examples are presented to illu strate the performance of the developed model and recursive algorithm.