Multimodal Data Fusion for Digital Monitoring of Railway Vehicle Wheelset Health Status

This study proposes a multimodal data fusion-based method for monitoring the health status of railway wheelsets by integrating vibration, acoustic, temperature and visual sensing information, and constructing an ‘edge-cloud’ collaborative computing architecture. The research innovatively adopts a deep fusion model with cross-modal attention mechanism, and realises high-precision identification and early warning of early failure of wheelsets. Practical engineering applications verify that the method significantly reduces the number of unscheduled wheelset replacements and maintenance costs, and shows strong robustness under severe operating conditions. The reliability and early warning capability of the system are greatly improved, providing a new paradigm for early warning of wheelset failures with obvious technical and economic benefits.