Variable step adaptive kurtogram method based on empirical wavelet transform for rolling bearing fault diagnosis

Envelope analysis is one of the most effective methods in rolling element bearing diagnostics, but finding an optomal frequency band (OFB) has been a challenge for a long time. In this study, a variable-step adaptive kurtogram (VSAK) method based on empirical wavelet transform (EWT) was proposed to select the OFB and apply it in fault detection of rolling element bearings. The VSAK method mainly contains signal reconstruction based on EWT and detection of the appropriate demodulation frequency band based on RK. A new weighted kurtosis-RK was applied in this VSAK method to reduce the influence of high gaussian noise. According to simulation and experimental studies, the results showed that, VSAK was superior over those traditional methods, such as FK and autogram, for the diagnosis of rolling element bearing. It was also concluded that, the sum of amplitude corresponding to the first five fault harmonic frequencies based on VSAK was larger than the sum of amplitude based on FK as well as autogram method.