Fault diagnosis of planetary gear based on entropy feature fusion of DTCWT and OKFDA

Planetary gears are often used in the key parts of the transmission systems of mechanical equipment, and faults are the main factors that determine the reliability of equipment operation. A fault diagnosis method for planetary gears based on the entropy feature fusion of dual-tree complex wavelet transform (DTCWT) and optimized kernel Fisher discriminant analysis (OKFDA) is proposed. The original vibration signal is decomposed by DTCWT, the frequency band signals are obtained, and the extraction models for the entropy features are built from multiple perspectives according to the definition of entropy theory. But the original entropy features, which are extracted from multiple perspectives, lead to excessive feature dimensions, and there are also many insensitive features that have small effects on the identification of the faults of planetary gears. Feature dimension reduction and sensitive feature extraction were achieved by OKFDA. The effectiveness of OKFDA and the extracted sensitive features were analyzed for the original features with different dimensions. Fault diagnosis for planetary gears can be achieved by analyzing sensitive features accurately.