It is inevitable for the collected gear vibration signals to contain some noises that can seriously affect the accuracy of the gear fault diagnosis.In the present work,an improved empirical wavelet transform(EWT)combined with wavelet threshold denoisingmethod was proposed.This method improved the traditional EWTfrequency spectrum segmentation method.First,the peak seeking algorithm was used to find the peak of the frequency spectrum,and then the smoothing algorithm was used to smooth the peak curve.The minimum points of the smooth curve were taken as the frequency spectrum segmentation boundaries,so that the divided filter banks were more accurate.The denoising results of simulated signals and experimental signals showed that the signal-to-noise ratio(SNR)of this method reaches 16.27,and the root mean square error(RMSE)reaches 7.54e-07.The proposed method had better denoising effectivity and higher robustness than the wavelet threshold,empirical mode decomposition(EMD)and traditional EWT.
Improved EWTwavelet thresholdempirical mode decomposition(EMD)denoising of the gear vibration signals
万方数据
维普
