Resnet with CBAM-SE for Bolt Fault Diagnosis in Simulated Noisy Industrial Environments

Abstract

This paper explores the use of a deep learning approach combining a ResNet50 model and CBAM-SE (Convolutional Block Attention Module with Squeeze-and-Excitation) for bolt fault diagnosis in a noisy industrial environment. The aim of the study is to improve the model's immunity to different noise conditions, covering fluid noise, impact noise, periodic noise and mixed noise. Experimental results show that the ResNet50+CBAM-SE model outperforms the conventional ResNet50 and CNN models in all noise cases, especially when dealing with complex noise disturbances.The addition of the CBAM-SE module enables the model to better focus on key features, thus improving its robustness and classification performance. This study demonstrates the potential of deep learning models with attention mechanisms for fault diagnosis, especially in industrial settings.