Fault detection method for the rolling bearings of metro vehicle based on RBF neural network and wavelet packet transform

To detect the rolling bearings fault of metro vehicle, a method combined wavelet packet with RBF neural network is proposed in this paper. Firstly, wavelet denoising is performed for the vibration signal to remove invalid signal. And then, energy characteristic vectors of the fault signals are extracted by wavelet packet decomposing to train the RBF neural network. Finally, the trained RBF neural network is used for fault classification. The diagnostic results show that the proposed method can be used to detect fault types of the metro vehicle rolling bearings precisely.