부산대학교 도서관

To improve the accuracy of the joint time-frequency analysis method for mechanical fault diagnosis, a gear fault diagnosis method, which combined with ensemble empirical mode decomposition (EEMD) and Choi-Williams distribution (CWD), is proposed in this paper. First, the collected faulty gear vibration signal is decomposed into a series of intrinsic modal components (IMFs) by EEMD, and IMFs are ranked in descending order of frequency. Then, the false components are removed and IMFs, which are "sensitive" to gear fault features, are selected based on the criteria of Kurtosis (K) and Cross-Correlation (CC). Finally, the selected IMFs are expressed in CWD Spectrogram, and the frequency and periodical impact features in time-frequency domain are combined to identify the gear fault characteristics. The applicability and effectiveness of this proposed method for gearbox gear fault diagnosis are verified by simulation and experimental analysis.