Abstract：To effectively extract the fault features of rolling bearings, an improved enhanced Kurtogram method based on harmonic wavelet packet decomposition was presented after analyzing the shortcoming of enhanced Kurtogram method based on binary wavelet packet decomposition. The node with the maximum kurtosis value could be selected after the improved enhanced Kurtogram of fault signal was computed, then used the harmonic wavelet packet coefficient of the optimal node to reconstruct the signal and analyzed the reconstructed signal by enhanced envelope spectrum. The fault type of rolling bearing could be judged by comparing the theoretical calculation value of fault characteristic frequency with spectral lines whose amplitude were obvious in enhanced envelope spectrum. Simulated inner fault signal and measured inner fault signal of rolling bearings were analyzed by this proposed method, and the diagnosis results showed that the new method was reliable and could meet the actual requirements.
2014, Vol. 33 
