Gear Fault Diagnosis based on EEMD and Choi-Williams Distribution

Abstract

In the process of gear fault feature extraction， empirical mode decomposition （EMD） has the problem of modal aliasing， intrinsic mode function（IMF） screening difficulty， and Wigner-Ville distribution（WVD） has cross-terms interference. So， a gear fault diagnosis method combining ensemble empirical mode decomposition（EEMD） decomposition and Choi-Williams distribution（CWD） is proposed. Firstly， the collected signal is decomposed by EEMD and decomposed into the combination of single component intrinsic mode functions. Then， the false components are removed and IMFs are screened through the correlation coefficient and Shannon entropy. Finally， the selected IMF components are expressed in CWD time frequency graph， and the gear fault features are extracted with the characteristics of frequency and isochronous impact exhibited in the time-frequency domain. Through simulation and experimental analysis， the applicability and effectiveness of this proposed method in gearbox gear fault diagnosis are verified.

Keywords

• Gear;Fault diagnosis;Ensemble empirical mode decomposition（EEMD）;Choi-Williams distribution （CWD）