The wavelet de-noising using neighboring coefficients and frequency slice wavelet transform (FSWT) were combined for weak fault time-frequency feature extraction of rolling bearing. Based on the analysis results of the vibration data of rolling element bearing's early weak fault

the strong background noise of rolling bearing can be decreased effectively by the wavelet de-nosing using neighboring coefficients method. Furthermore

the de-noised signal was handled by the FSWT method and better time-frequency feature extraction result was obtained compared with the method using FSWT directly

so the effectiveness of the proposed method was verified. Besides

the advantages of the proposed method were also verified by comparing with other time-frequency method such as spectral kurtosis.