GAF-Wavelet Fusion and SENet-Driven Domain Adaptation for Bearing Fault Diagnosis

Bearing fault diagnosis is crucial to ensure the safe operation of rotating machinery equipment. However, the existing methods based on deep transfer learning still have some problems in cross condition scenarios. For example, the difference in the feature distribution between the source domain and the target domain during the migration process leads to a decline in the generalization performance of the model and a decline in the accuracy of fault identification. For this reason, this paper proposes a data processing method based on the fusion of Gram Angle Field (GAF) and Continuous Wavelet Transform (CWT), and constructs a fusion feature map with high discrimination. In the feature extraction phase, the SENet attention mechanism module is introduced into Alexnet model to dynamically optimize the channel feature weight, and a joint Correlation Alignment (CORAL) and Deep Adaptation Networks (DAN) adaptive weight strategy is designed to significantly improve the ability of domain adaptation and domain alignment. The experimental results show that the proposed method improves the accuracy of fault diagnosis on the CWRU bearing dataset.