A New Identification Method for Hammerstein Model Based on PSO

This paper presents a new approach of structure identification and parameter estimation for Hammerstein Model by using particle swarm optimization (PSO). The average square error criterion (ASE) has been proposed to decrease computation and obtain the true optimal structure effectively. Meanwhile, the modified identification algorithm is always converging by backward algorithm, and thus obtaining a high precision for the parameter estimation. Simulation results indicate that the ASE is an efficient order selection criterion, but Akaike´s information criterion (AIC) and minimum description length (MDL) are not good for order selection and parameter estimation in Hammerstein model