A fault diagnosis approach for roll bearing based on wavelet-SOFM network

A novel method of pattern recognition and fault diagnosis in roll bearing based on the wavelet-neural network is proposed according to the frequency spectrum characteristics of vibration signal. Based on the advantage of multi-dimensional multi-scaling decomposition of wavelet packets, the abrupt change information can be obtained and the features related to the fault of roll bearing is extracted through the decomposing and reconstruction of the vibration sign of the roll bearing. The extract features are inputted into SOFM to realize the automatic classification of the fault. The trained SOFM can be used to the online state monitor and real-time fault diagnosis of roll bearing. The feasibility of this novel method is proved by the simulation results.