Research on gear and bearing fault diagnosis based on deep residual shrinkage network

In modern industry, the regular operation of air compressors and pumps is essential, and the state of their key components bearings and gears is directly impacts the operational efficiency and the equipment service life. Advancements in deep learning have propelled deep learning-based approaches to the forefront of research, but the current fault diagnosis models have insufficient generalization ability when handling cross-dataset applications and cross-condition scenarios. In this study, a fault diagnosis model for bearings and gears based on a deep residual shrinkage network is constructed, and the Adam optimization algorithm is used to improve learning efficiency. Through the training and verification of gear and bearing data sets, the test results show that the model achieves a fault diagnosis accuracy of 99.2%. The model can automatically obtain the noise reduction threshold, enhance the fault feature extraction, suppress the noise interference, improve the model's generalization capability, and provide an effective solution for the fault diagnosis of bearings and gears.