Multi-Scale Convolutional Neural Network Fault Diagnosis Based on Attention Mechanism

With the development of artificial intelligence, the method of intelligent fault diagnosis is of great significance for industrial production. Aiming at the adaptability of rolling bearings under different working conditions under different fault diameters, an intelligent fault diagnosis method based on the combination of multi-scale convolutional neural network and attention mechanism is proposed. The structure of multi-scale convolution is used to extract features of different dimensions and significantly improved feature extraction capabilities. Besides, the attention mechanism is used to give different weights to data features and pay attention to its important characteristic information. In order to verify the performance of the Multi-scale Convolutional Neural Network-Convolutional Block Attention Module (MSCNN-CBAM) model, the bearing bench dataset of Case Western Reserve University was used for verification. The results show that the approach is efficient in diagnosis and experimental results show that compared with other models, the MSCNN-CBAM model can capture high-quality features more effectively and show strong robustness under different working conditions. The model has a good fault diagnosis capabilities and achieved a classification rate of more than 99%.