Aiming at the difficulties of industrial data acquisition and the limited ability of sparse data feature extraction,a rolling bearing fault diagnosis model based on Mel spectral data enhancement and ResNet network was proposed.In order to increase the diversity of training sample data,Mel spectral data enhancement technology was introduced,and the generalisation ability of the model was improved,so it could better adapt to variety rolling bearing fault situations.Through the stacking depth of the residual units of the ResNet network,the complex fault feature information could be captured,and the fault mode of rolling bearings was identified effectively.Finally,the rolling bearing fault data collected at the experimental site were validated and evaluated. The results show that the diagnostic accuracy of the proposed model on the enhanced dataset is as high as 99.83%,which is 1.39% higher than the original dataset.Compared with other methods,the model achieves significant improvements in accuracy and robustness,and is able to identify different fault types of rolling bearings more accurately.