Skip to Main Content
The incipient fault diagnosis of rolling bearings is the technical prerequisite for safe production and avoiding heavy accidents. In this paper, an intelligent incipient fault diagnosis method is developed, using hybrid wavelet and neural networks. The high frequency noises in the vibration signals from rolling bearings are first eliminated by the adaptive wavelet de-noising, then the purified signals are transformed by wavelet-packet to extract the energy feature in each subband to form the fault feature vectors. The mapping relationship between fault features and fault modes is set up by a wavelet neural network. Experiments show the above method is reliable in the incipient fault diagnosis of rolling bearings.