Intelligent Fault Diagnosis of Gearbox Based on Residual Network and Data Enhancement

The gearbox in the mechanical transmission system runs for a long time under complex working conditions, and its internal parts will inevitably fail. In order to ensure the safe and stable operation of the mechanical system, it is necessary to monitor the condition of the key parts of the gearbox and effectively judge the fault state. At present, the fault diagnosis model based on machine learning has developed rapidly, but the model effect often depends on a large number of high-quality training data samples, which is difficult for the fault diagnosis task, and the fault data generated by the actual operating conditions is limited. And there is often a class imbalance phenomenon. Based on this, this paper uses the data enhancement strategy to expand the class unbalanced data set, and combines the residual neural network to achieve fault feature extraction and accurate diagnosis. The experimental results can prove the effectiveness of the proposed method.