A Bearing Fault Diagnosis Model Combining Multi-Scale Features and Attention Mechanisms

Bearing fault diagnosis is of great significance in industrial equipment maintenance. Improving the accuracy and robustness of the diagnosis model is the focus of current research. To this end, a bearing fault diagnosis model (MS-TACBAM) combining a multi-scale convolutional neural network, a channel and spatial attention mechanism (CBAM), and a temporal attention transformer is proposed. The method first extracts feature information under different receptive fields through a multi-scale convolution module, and uses a CBAM module to enhance the attention of key features. Subsequently, the temporal attention transformer is introduced to model time series signals to capture long-term dependencies and improve the ability to recognize complex fault modes. The experiment is verified using the Case Western Reserve University bearing fault dataset. The results show that the method is superior to the comparison model in terms of diagnostic accuracy, feature extraction capability, and computational efficiency, and shows good robustness under different load conditions.