An Improved Autogram and its Application in Bearing Fault Diagnosis

Envelope analysis is one of the most advantageous methods in rolling element bearing (REB) diagnosis, but the determination of the informative frequency band (IFB) is always a huge challenge. In order to reduce the influence of discrete frequency components and strong background noise to choose IFB, a new improved Autogram is provided. Firstly, the Cepstrum pre-whitening (CPW) is taken as a preprocessing method to reduce the interference from discrete frequency components in the signal. Then, the Autogram is used to decompose the pre-whitening signal and the optimal IFB is selected for a further demodulation. The simulated signals and a set of experiment signals are taken to verify its validity. The results indicates that this method can successfully address the interference generated by discrete frequency and strong background noise, and effectively extract the weak fault features of bearing.