Feature Extraction of Bearing Faults based on a Novel Index of Cepstrum

Bearing is the common component used in modem mechanical equipment, its fault feature extraction is of importance to the safe operation. A new feature extraction index named pseudo distance of cepstrum edge ( PDCE) is proposed in this paper, and the detailed process is discussed. Four principles of determining the diacritical standard is summed up. In order to validate the capability of this new method, 8000 points of vibration data representing four different operational status are adopted, which are all from practical bearing fault experiment. Numerical simulation experiment including a test simulation analysis is carried out, results testify the usability of present index.