Automatic selection of frequency-band based on wavelet for gear fault diagnosis

In order to realize on-line monitor and diagnosis of the fault of the gearbox of a wind driven generator a method of automatic selection frequency-band based on wavelet analysis and kurtosis rule has been put forward. Firstly, a vibration signal is multiple discomposed by using wavelet transformation, and then the kurtosis for wavelet coefficients of different frequency bands is computed and compared. Secondly, the frequency band containing fault frequency can be obtained according to the criterions of maximum kurtosis. Finally, the fault frequency is extracted by analyzing envelope spectrum of the frequency band selected. This method is capable of extracting fault frequency of broken teeth exactly. Through simulation experiment, analysis of theory and experiment of broken teeth faults, the results show that automatic selection of frequency-band based on wavelet is an effective method for online diagnosis of a gearbox.