Identification of Wiener system with time delay using correlation analysis and swarm intelligence methods

Abstract

In this paper, a novel identification method is addressed for a class of nonlinear Wiener systems subject to time delay, and this identification problem involves the estimation of delay time, transfer function model parameters and neural fuzzy model parameters. Aiming to identify separately the linear and nonlinear blocks, the separable signals are introduced. First, the correlation characteristics of separable signals through the Wiener system are analyzed, then the correlation analysis technique is applied to calculate the unknown parameters involving time delay and transfer function model. Moreover, in the neural fuzzy model parameters estimation, we first calculate the center and width of the neural fuzzy model. Then, to improve global search mechanism ability and converge speed of particle swarm optimization method, the improved particle swarm optimization and cuckoo search techniques are introduced to figure out the weight of the neural fuzzy model, which obtains good global search ability and convergence speed. The simulation comparison results in numerical case and nonlinear process are presented to verify that the feasibility of the Wiener system identification.