基于雷视融合的高铁周界人员入侵检测方法OA
Personnel Intrusion Detection Method for High-Speed Railway Perimeter Based on Radar-Camera Fusion
[目的]近年来,高铁行车安全问题依然严峻,人员或异物入侵周界造成铁路安全事件时有发生,严重影响列车运行秩序.高铁周界要求能够在强降雨、浓雾等较为恶劣天气条件以及全时段进行防护,因毫米波雷达具备全天候、全天时精准监测的独特优势,成为恶劣天气条件下的周界监测首选传感器.高速铁路周界入侵监测具有长线性、气候多样、地形复杂等特点,并且监测精度要求高.因此,提出基于毫米波雷达和相机融合的人员入侵监测方法.[方法]首先,通过毫米波雷达和球型相机联动的方法,实现对远距离目标的聚焦放大,提升远距离目标的成像效果.其次,为提高恶劣天气条件下视频识别效果,提出一种基于天气判识的目标检测方法,通过YOLOv9 网络进行天气判识,并将天气判识结果与目标检测网络的权重系数关联,提升恶劣天气条件下视频识别的准确率.然后,提出一种基于目标跟踪的融合检测算法,在图像失真的情况下,通过毫米波雷达的连续跟踪能力输出报警结果,提升融合识别的准确率.[结果]在模拟恶劣天气的试验场进行测试 1 540 次,其中雨雾天气条件下的人员入侵测试 1 320 次,无入侵的干扰测试 220 次,结果表明,该方法能够满足 30mm/h 及以下降雨、强浓雾天气200 m 以内人员入侵报警,漏报率 0％,误报率 3.2％.[结论]通过在渝昆高铁开展正线试验,均能正确报警,验证了基于雷视融合的方法能够满足周界入侵检测的需求.
[Objective]In recent years,safety issues in high-speed railway operation remain severe.Railway incidents caused by intrusion of personnel or foreign objects into the perimeter occur frequently,seriously disrupting the order of train operation.The perimeter of high-speed railways requires protection under severe weather conditions such as heavy rainfall and dense fog,as well as throughout the entire period.Due to the unique advantages of all-weather and all-time precise monitoring,millimeter-wave radar has become the preferred sensor for perimeter monitoring under harsh weather conditions.The perimeter intrusion monitoring of High-speed Railways is characterized by long distance,diverse climatic conditions,complex terrain,and high monitoring accuracy requirements.Therefore,a personnel intrusion monitoring method is proposed based on the fusion of millimeter-wave radar and camera.[Methods]First,a method combining millimeter-wave radar and dome camera was utilized to achieve the focused magnification of long-distance targets,thereby improving the imaging performance of such targets.Then,to improve the performance of video recognition under harsh weather conditions,a target detection method based on weather identification was proposed.This method used the YOLOv9 network for weather identification,and correlated the identification results with the weight coefficients of the target detection network,thereby enhancing the accuracy of video recognition under adverse weather conditions.Moreover,a fusion detection algorithm based on target tracking was proposed,which generated alarm results using the continuous tracking capability of millimeter-wave radar in the case of image distortion,thereby improving the accuracy of fusion recognition.[Results]A total of 1 540 tests were conducted at a test site simulating harsh weather conditions,including 1 320 personnel intrusion tests under rainy and foggy conditions and 220 interference tests without intrusion.The results showed that this method could meet the requirements of personnel intrusion alarm within 200 meters under rainfall of 30 mm/h or less,as well as under heavy fog weather conditions,with a missed alarm rate of 0％and a false alarm rate of 3.2％.[Conclusion]Mainline tests conducted on the Chongqing-Kunming high-speed railway show that all intrusion events are correctly alarmed,demonstrating that the proposed method based on the fusion of radar and camera can meet the requirements of perimeter intrusion detection.
郭鹏跃;史天运;王瑞;马祯;张万鹏
中国铁道科学研究院集团有限公司电子计算技术研究所,北京 100081中国铁道科学研究院集团有限公司科技和信息化部,北京 100081中国铁道科学研究院集团有限公司电子计算技术研究所,北京 100081中国铁道科学研究院集团有限公司电子计算技术研究所,北京 100081中国铁道科学研究院集团有限公司电子计算技术研究所,北京 100081
交通工程
高速铁路行车安全周界入侵雷视融合人员检测零漏报
high-speed railwayoperation safetyperimeter intrusionradar-camera fusionpersonnel detectionzero missed alarm
《铁道标准设计》 2026 (6)
184-191,8
国家自然科学基金项目(U2268617)国家重点研发计划项目(2022YFB4300604)
评论