Abstract:As the most important part of flying wheel, the characteristics of bearing assembly influence life of flying wheel. So it is very important for bearing assembly to diagnose incipient faults. At the initial stage of bearing assembly failure, its fault characteristics are relatively weak, and the fault signal is easily submerged in the background noise signal, which brings great difficulties to the early fault judgment and positioning of bearing assembly. Based on the vibration test method of fast envelope kurtogram, the vibration signal of the collected bearing assembly is denoised by wavelet to improve the signal-to-noise ratio of the special diagnosis signal firstly. Then the full-frequency kurtosis analysis of the vibration signal is carried out. The incipient fault of bearing assembly is diagnosed and located exactly in order to get the fault frequency, which will offer an effective method for screening bearing assembly.