A hybrid method based on conjugate gradient trained neural network and differential evolution for non linear systems identification

Abstract

A hybrid method based on Differential Evolution and Neural Network training algorithms is presented in this paper for improving the performance of neural network in the non linear system identification. For this purpose, the local optimization algorithm of conjugate gradients (CG) is combined with the differential evolution algorithm (DE), which is a population-based stochastic global search method, to yield a computationally efficient algorithm for training multilayer perceptron networks for nonlinear system identification. After, a series of simulation studies of our method on the different nonlinear systems it has been confirmed that the proposed CG+DE algorithm has yielded better identification results in terms of time of convergence and less identification error.