Due to the interference of wheel-rail excitation,wheel-set bearing fault characteristics are easily submerged by background noise.The time-varying fault characteristics in variable speed mode further aggravate the difficulty of extrac-tion.For this purpose,a fault feature extraction method for train wheelset bearings was proposed based on the fast spec-tral average kurtogram.By dividing spectrum into several sub-bands according to the spectral trend,and using the Meyer wavelet to construct a band-pass filter,the average kurtosis of each narrowband filtered signal was calculated and a fast spectral average kurtogram was constructed.Based on the selection of the optimal frequency band for demodulation to eliminate the influence of background noise,the iterative generalized demodulation of the filtered signal was carried out by phase function to solve the spectrum ambiguity caused by time-varying fault impact intervals.The simulation and ex-perimental analysis results show that the proposed method effectively overcomes the interference of strong background noise and time-varying fault features,and provides a new solution strategy for wheel-set bearing fault feature extraction under variable speed conditions.
维普
万方数据