Fault diagnosis model for railway signalling equipment using deep learning techniques

A hybrid deep transfer learning-assisted fault diagnosis model (HDTL-FLM) was presented for railway signalling equipment, addressing the challenge of accurately diagnosing and predicting irregularities in this critical transportation component. The model incorporates a feature extraction method based on variational mode decomposition and leverages multi-scale signal factors for information capture during transfer learning processing. The use of a transfer learning backend enables alignment of cross-domain features, allowing the HDTL-FLM to detect and diagnose faults under various working conditions. Experimental results demonstrate that the proposed model enhances diagnostic accuracy, fault prediction ratio, and reduces error rate compared to existing models, making it a promising solution for maintaining safe and efficient railway operations.