Design and Application Verification of UAV Intelligent Inspection System for High-Speed Railway Bridges

To address the limitations of traditional manual inspection methods for high-speed railway bridges, this paper proposes an integrated UAV-based intelligent inspection system. The system is designed to enhance inspection efficiency, accuracy, and standardization by combining automated flight control, image data acquisition, defect detection, and risk evaluation. It adopts a modular, six-layer architecture encompassing perception, communication, computation, intelligent analysis, platform coordination, and visualization. Deep learning models are employed to localize key structural components and detect various defect types, including small-scale anomalies and complex surface damage. A quantitative evaluation module assigns risk levels to detected defects and supports maintenance prioritization based on severity and component importance. Field validations were conducted on steel truss bridges and urban noise barriers, demonstrating the system’s adaptability to different bridge types and operational environments. The results confirm the system’s capability to deliver structured and actionable outputs, thereby supporting data-driven maintenance decisions. This study contributes a scalable and generalizable solution for intelligent railway infrastructure inspection, offering a foundation for further integration with predictive maintenance and digital asset management platforms.