Resampling Technique-Based Demodulation Analysis for Planet Bearing Cage Fault Diagnosis Under Nonstationary Conditions

Abstract

Vibration signals of the planetary transmission system usually contain excitation responses from different structures. The fault characteristic of the planet bearing cage is weak, which is often overwhelmed by strong noise and other irrelevant vibration components. Therefore, fault feature extraction of the planet bearing cage is a challenging topic. In this paper, a novel fault feature extraction method is proposed to extract weak fault features of the planet bearing cage. The general parameterized time-frequency transform is employed to accurately extract instantaneous rotational speed information from the planet bearing vibration signal for resampling. It can obtain the accurate rotational speed result even though under power frequency interference conditions. The influence of the deterministic components excited by the gear can be reduced by using the spectral pre-whitening and the interference of noise and random impact can be reduced in the selection of the optimal resonance frequency band by using autogram analysis. The fault characteristic components of the planet bearing cage can be extracted from the square envelop spectrum. The neighborhood power density ratio index is further constructed to distinguish the health bearing state from the cage fault bearing state and evaluate the degree of faults. Experiments designed in a planetary transmission system test rig of an armored vehicle have verified the potential and effectiveness of the proposed fault feature extraction method in exhibiting planet bearing cage fault features.