Bearing Fault Diagnosis Based on One-Dimensional Convolution Network and Residual Training

Bearings are an important part of large industrial production equipment, and their damage is often accompanied by significant economic losses and personal safety issues. Algorithms based on the traditional machine learning, the recognition rate is not high, and the data preprocessing has great influence on the result. Models based on deep learning, which is relatively simple, and the shallow features extracted include noise. Aiming at the above problems, this paper proposes an end-to-end deep one-dimensional convolution residual network (DOD-CRN) fault diagnosis model. Training on the original vibration signal, and the feature extractor is constructed according to the characteristics of the data, which included the extraction of shallow features and high-level features. Experiments show that DOD-CRN has 98.3% fault recognition rate under complex fault signals, which is higher than other algorithms and can effectively detect faults.