Recursive maximum likelihood method for the identification of Hammerstein ARMAX system

Abstract

Identification of Hammerstein nonlinear models has received much attention due to its ability to describe a wide variety of nonlinear systems. This paper considers the parameter estimation problem of ARMAX models for the Hammerstein systems. The recursive maximum likelihood method, which can be applied to online identification and occupies small memory capacity, is proposed to deal with the problem in here. It is an approximation of the maximum likelihood method. The parameters of the linear and nonlinear parts of the Hammerstein model and the noise model can be directly obtained without using the overparameterization technique. Finally, the proposed method is applied to a classic Hammerstein ARMAX system and is compared with RLS method in detail. The research results show the effectiveness of the proposed method.