the deficiency of full classification and obvious distinguish by using Self-organizing Feature Map(SOM) network is analyzed and a SOM Neural Network algorithm based on Particle Swarm Optimization(PSO) is presented.Being easy to realize
fast to convergence
the parameters are optimized by the algorithm and vibration faults diagnosis of the shaft system is simulated by the optimized SOM Neural Network.The diagnostic results of the instance show that it has better classifying results than that of ROM Neural Network.