Nonlinear Identification and Control of the Hammerstein System with Application to pH Neutralization Process

Abstract

This paper deals with the nonlinear identification modeling and control based on the Hammerstein structure for the pH neutralization process. The Hammerstein structure is composed of a static nonlinear part approximated by Gaussian basis function-based neural network and a dynamical linear part modeled by transfer function, and the parameters separation identification of the Hammerstein system is realized by using the characteristics that the binary signal does not excite the static nonlinear system. As a consequence, the linear part parameter are identified according to least square method. Furthermore, the parameters of Gaussian basis function based feedforward neural network are learned using gradient descent method and error back propagation with the help of measurable random signals. Simulation results demonstrate that the proposed Hammerstein structure can effectively represent the pH neutralization process, and the proposed algorithm can acquire good tracking performance.