Abstract: The gearbox gear wear tend to result in impulse response in vibration signals, and monitoring impulse response in the rotation cycle, which can achieve fault diagnosis. In order to improve monitoring and diagnosis effects by the visualization, distribution method of the polar digram angle frequency(DPDAF) was introduced in the study. Gearbox vibration signals were denoised by continuous wavelet transform, and then transformed into DPDAF, which can clearly exhibite the impulse response signal differences with the six rotation cycles in the different wear conditions. Six rotation cycle energy were extracted as the feature vectors of gearbox gear wear fault, which were used to train BP neural network for fault pattern recognition. Test results showed that applying DPDAF and BP neural network to gearbox gear wear fault diagnosis was feasible and effective. The results provide a reference for the engineering applications of the polar angle frequency representation in the gearbox condition monitoring and fault diagnosis.