Application of neural network integration in fault diagnosis

Abstract

According to the generation methods of individual neural network and the methods of generating conclusions from integrated neural network, an effective neural network integration system can be constructed. An optimization method for neural network integration is proposed. In the generation of individuals in the network integration, a variety of genetic algorithms and particle swarm optimization algorithm are used to train individual networks, thus to improve the precision of network members and reduce the correlation among the network members; in the conclusion generation, weight of the individual neural network is dynamically determined. The simulation results show that the effectiveness and feasibility of the method in fault diagnosis. © Xiong et al.