Motor Bearing Fault Diagnosis Based on Phase Space Reconstruction and IGAM-Swin Transformer

To address the low accuracy in motor bearing fault diagnosis caused by traditional methods' poor handling of nonlinear signals, we propose a fault diagnosis method based on phase space reconstruction (PSR) and IGAM-Swin Transformer. First, PSR is applied to restore the system's dynamic characteristics, converting the one-dimensional vibration signal into a two-dimensional phase diagram. This diagram is then input into the IGAM-Swin Transformer model, which integrates an optimized global attention mechanism (IGAM) to enhance feature extraction. The model is further optimized through transfer learning with pre-trained parameters. Finally, the model classifies different fault types. Simulation results show that the proposed method achieves a fault diagnosis accuracy of 99.78%, outperforming comparison methods. Additionally, the IGAMSwin Transformer improves classification performance over the standard Swin Transformer. The results confirm the method's effectiveness and demonstrate that nonlinear signals can be processed with high accuracy using phase space reconstruction and the IGAM-Swin Transformer for motor bearing fault diagnosis.