基于YOLOv4目标检测算法的城市轨道交通刚性接触网悬挂状态智能检测系统的设计与应用
马志鹏1袁万全2周于翔2魏志恒2
Design and Application of Intelligent Detection System for Urban Rail Transit Rigid Catenary Suspension Status Based on YOLOv4 Object Detection Algorithm
MA ZhipengYUAN WanquanZHOU YuxiangWEI Zhiheng
-
作者信息:1.中国铁道科学研究院， 100081， 北京
2.中国铁道科学研究院集团有限公司城市轨道交通中心， 100081，北京
-
Affiliation:China Academy of Railway Sciences, 100081, Beijing, China
-
关键词:
-
Key words:
-
DOI:10.16037/j.1007-869x.2023.10.027
-
中图分类号/CLCN:U226.8+1
-
栏目/Col:应用技术
摘要:
目的：为提高城市轨道交通刚性接触网典型病害的检测效率与精度，降低成本，保障线路运营安全，需对刚性接触网悬挂状态检测方法进行进一步研究。方法：介绍了YOLOv4目标检测算法，阐述了该算法的优点；提出了基于YOLOv4目标检测算法的刚性接触网悬挂状态智能检测系统，详细介绍了该系统的架构组成及技术参数；将基于YOLOv4目标检测算法的刚性接触网悬挂状态智能检测系统应用于北京地铁某线路刚性接触网悬挂状态的检测中。结果及结论：基于YOLOv4的刚性接触网悬挂状态智能检测系统实现了对刚性接触网悬挂状态的快速巡检及全面高清成像，以及对接触网零部件悬挂状态病害的智能分析及数据管理。该系统在北京地铁某线路上取得了良好的工程应用效果，可在车载运行状态下完成对接触网悬挂状态的智能检测。
Abstracts:
Objective: To enhance the detection efficiency and accuracy of typical defects in the rigid catenary of urban rail transit, reduce costs and ensure operational safety, further research on the detection method of rigid catenary suspension status is necessary. Method: The YOLOv4 (You Only Look Once version 4) object detection algorithm is introduced, and its advantages are elaborated. An intelligent detection system for rigid catenary suspension status based on YOLOv4 object detection algorithm is proposed, providing a detailed description of the system architectural composition and the required technical parameters. This is applied to the detection of rigid catenary suspension status of a Beijing Subway line. Result & Conclusion: The YOLOv4based intelligent detection system for rigid catenary suspension status realizes rapid inspection and comprehensive highdefinition imaging of the suspension status, as well as intelligent analysis and data management of defects in the catenary component suspension status. The system has demonstrated positive engineering application results on a Beijing Subway line, enabling intelligent detection of the catenary suspension status during onboard operation.
过刊浏览
资料下载
友情链接