Fault Diagnosis Method of Gearbox Based on Dual-Tree Complex Wavelet Packet Transform and IBA-BP

In view of the non-linear and non-stationary characteristics of gearbox fault signal, BP neural network has the problem of low fault recognition rate of gearbox. This paper proposed a gearbox fault diagnosis method based on dual-tree complex wavelet packet transform and BP neural network optimized by improved bat algorithm (IBA). Firstly, the gearbox vibration signal is decomposed and reconstructed in three layers through dual-tree complex wavelet packet and the energy features are extracted in the reconstructed signal, Then feature samples are trained and fault classification identified by BA-BP. The initial weights and biases of BP were optimized by improved bat algorithm. Experiments show that the fault diagnosis method based on DT-CWPT and IBA-BP can identify the fault more effectively.