As the operational length of railways in China continues to expand,traditional patrol inspection methods and modes have increasingly struggled to meet the growing requirements of railway operations for safety and bene...As the operational length of railways in China continues to expand,traditional patrol inspection methods and modes have increasingly struggled to meet the growing requirements of railway operations for safety and benefits.Based on an analysis of railway infrastructure patrol inspection requirements,this paper presents the development of a space-air-ground integrated intelligent patrol inspection technology system that leverages emerging technologies,including UAVs and satellite remote sensing.It also proposes an application solution and implementation path,along with the design of a componentized and plug-in system framework.Ablation experiments on the plug-in-based algorithm for intelligent detection of color steel tile hazards demonstrated the proposed BINDeepLabV3+algorithm's strong applicability to both satellite and UAV images,achieving an mIoU of 89.11%and an mPA of 94.83%.These results confirm the implementability of the plugin approach.Demonstration applications conducted on a selected section of the Beijing-Shanghai High Speed Railway validated the feasibility and advancement of the proposed system.The space-air-ground multi-source patrol inspection approach enables complementarity and collaboration,facilitating the timely and effective identification of a greater number of safety hazards.This approach provides a multi-faceted,full-process,one-stop,intelligent,and lean application model for railway infrastructure patrol inspections.展开更多
文摘As the operational length of railways in China continues to expand,traditional patrol inspection methods and modes have increasingly struggled to meet the growing requirements of railway operations for safety and benefits.Based on an analysis of railway infrastructure patrol inspection requirements,this paper presents the development of a space-air-ground integrated intelligent patrol inspection technology system that leverages emerging technologies,including UAVs and satellite remote sensing.It also proposes an application solution and implementation path,along with the design of a componentized and plug-in system framework.Ablation experiments on the plug-in-based algorithm for intelligent detection of color steel tile hazards demonstrated the proposed BINDeepLabV3+algorithm's strong applicability to both satellite and UAV images,achieving an mIoU of 89.11%and an mPA of 94.83%.These results confirm the implementability of the plugin approach.Demonstration applications conducted on a selected section of the Beijing-Shanghai High Speed Railway validated the feasibility and advancement of the proposed system.The space-air-ground multi-source patrol inspection approach enables complementarity and collaboration,facilitating the timely and effective identification of a greater number of safety hazards.This approach provides a multi-faceted,full-process,one-stop,intelligent,and lean application model for railway infrastructure patrol inspections.
