Fault Diagnosis Method for Bearing and Gear Based on Deep Learning Waveform Image Recognition

摘要

A fault diagnosis method for bearing and gear based on deep learning waveform image recognition is proposed in this paper. The method can extract valid fault features automatically from time and frequency domain waveform curves of raw vibration signal and complete fault diagnosis. The training method for CNN model based on deep learning combined with transfer learning is designed, which needs only a small amount of faults sample data. The fault diagnosis algorithm and steps are given. The method is tested with a set of time and frequency domain vibration waveform data of gears and bearings in normal and different fault states. The results show that the proposed method has high recognition accuracies for the different types of faults and has good applicability.