Asynchronous input gear damage diagnosis using time averaging and wavelet filtering

Vibration signals are often used for fault diagnosis in mechanical systems because they are containing dynamic information of mechanical elements. Vibration signals from a gearbox are usually noisy and the signal-to-noise ratio (SNR) is so low that feature extraction of signal components is very difficult, especially in practical situations. One of the solutions to this problem is applying signal time-averaging techniques in time domain for signal denoising, but using this method is only possible when gearbox input shaft rotation is constant or synchronous. In this paper, a new noise canceling method, based on time-averaging method for asynchronous input, is developed, and then complex Morlet wavelet is implemented for feature extraction and diagnosis of different kind of local gear damages. The complex Morlet wavelet, used in this work, is adaptive because the parameters are not fixed. The proposed method is implemented on a simulated signal and real test rig of Yahama motorcycle gearbox. Both simulation and experimental results have proved that the method is very promising in analysis of the signal and fault diagnosis of gearbox.