Phase Information at Tooth Mesh Frequency for Gear Crack Diagnosis

This paper proposes a new method of frequency spectrum analysis of gear mesh vibration signals for extracting phase information directly related to individual tooth conditions. The modulation effect of a gear crack on gear mesh vibration is theoretically analyzed and a crack identification criterion is given. Resampling-based synchronous time averaging technique is employed to enhance gear mesh vibration. The enhanced vibration signal is then divided into segments corresponding to individual tooth mesh durations, and Discrete Fourier Transform is performed to each signal segment respectively, giving a series of tooth-related amplitude and phase spectra. A crack in a gear is identified through tooth-by-tooth comparison of the spectrum components of the same frequency from the spectrum series. Experimental results of a pair of bevel gears show that the phase information is more valuable for gear crack detection and the proposed approach makes the diagnosis easier, more effective and reliable compared with other methods.