Research on fault diagnosis based on BP neural network optimized by chaos ant colony algorithm

Abstract

In view of shortcomings of BP neural network, which is slow to converge and tends to trap in local optimum when applied in fault diagnosis, an approach for fault diagnosis based on BP neural network optimized by chaos ant colony algorithm is proposed. Mathematical model of chaos ant colony algorithm is created. Real-coded method is adopted and the weights and thresholds of BP neural network are taken as ant space position searched by chaos ant colony algorithm to train BP neural network. Training result of chaos ant colony algorithm is compared with that of conventional BP algorithm and from both results it is can be seen that chaos ant colony algorithm can overcome the shortcomings of BP algorithm. It is proved that mathematical model of chaos ant colony algorithm is correct and optimization method is valid through experimental simulation for machinery fault diagnosis of mine ventilator. © 2010 Springer-Verlag.