Feature extraction of transient signal based on double layer auditory nerve oscillator network

Rolling bearing plays an important role in rotary machines. In rotating machine fault diagnosis, calculating the characteristic frequency of the signal is a useful method. Human auditory nerve system can integrate the binaural information through the mechanism of neurons oscillation and delivery of oscillation. In view of the aspects mentioned above, to simulate the operating mechanism of human binaural auditory system, a double layer auditory nerve oscillator network (DLNON) model whose inputs are two features of the signal, is proposed for features extraction and faults diagnosis. It includes the basement membrane, inner hair cells, feature extraction, oscillation network 1(ON1), oscillation network 2(ON2) and oscillation network 3(ON3). After that, calculating the oscillation period of each oscillating element in the second layer network and according to the size of the oscillation period we can determine whether there is a transient impact component. The performance of DLNON model is evaluated by experiments. The results show that the model can effectively extract fault features, and distinguish fault types.