Next-Gen AIoT-Driven Predictive Maintenance and Health Management System for High-Speed Train Wheel and Axle Systems

This paper proposes an AIoT-driven predictive maintenance and health management system specifically designed for high-speed train wheel and axle systems. By integrating advanced artificial intelligence techniques—such as Multi-Scale Transformer (MST) for accurate wheel wear prediction and Deep Wavelet Network (DWN) for precise axle stress analysis—within a comprehensive Internet of Things framework, the system enhances operational efficiency and safety. To address challenges like data quality, model interpretability, and real-time processing constraints, the research incorporates edge computing solutions to improve responsiveness. Experimental results demonstrate a 22% improvement in predictive accuracy for wheel wear and a 27.5% reduction in mean absolute error for stress prediction compared to traditional methods. Additionally, anomaly detection performance achieved an F1-score of 0.90, outperforming existing models by 26.8%. These findings validate the system's effectiveness and lay the groundwork for future advancements in intelligent railway systems, emphasizing data fusion, interpretable AI, and real-time processing capabilities.