Bearing fault diagnosis using Wavelet analysis

One-dimensional discrete wavelet transform is used to process the bearing fault signal in this paper. Firstly, the bearing fault data is decomposed to multi-layer. Then the fault feature signal is reconstructed. In order to detect the bearing failure and determine the area of it, the reconstructed signal is processed by the Hilbert transform demodulation and spectrum refining. The results show that the frequency of failure point matches well with theoretical one using this method. This method is simple and reliable and thus provides a scientific method for early warning and exclusion of failure.