Identification of Nonlinear Time-delay System Using Multi-dimensional Taylor Network Model

Abstract

Combining multi-dimensional Taylor network (MTN) model with improved conjugate gradient (ICG) method, named ICG-MTN, is proposed for identification of nonlinear time-delay system in this paper. MTN is regarded as the identification model relying on its characteristic of strong approximation, and ICG method is regarded as the learning algorithm of MTN. Meanwhile, back propagation neural network (BPNN) is regarded as the method of comparison. By the experimental results, the nonlinear time-delay system can be identified effectively by the proposed method, and the effectiveness is better than the BPNN.