东南大学学报（自然科学版）2026，Vol.56Issue(10)：1392-1403,12.DOI:10.3969/j.issn.1001-0505.2026.10.003
基于YOLOv11-CL和GPU并行RTM的城市道路路基隐蔽病害快速识别定位
Identification location of hidden defects in urban roads based on YOLOv11-CL and GPU parallel RTM
摘要
Abstract
In response to the problems of limited accuracy in detecting and identifying various underground hidden defects in urban road using ground penetrating radar(GPR)and the difficulty in accurately characteriz-ing the location and morphology of these defects,a rapid identification and location technology for hidden de-fects based on YOLOv11-CGAFusion-LSDECD(YOLOv11-CL)and reverse time migration(RTM)was pro-posed.First,an intelligent identification model for void,loose and pipelines based on the improved YO-LOv11-CL was constructed to achieve accurate identification of various hidden defects in complex urban road subgrades.Then,an RTM imaging location algorithm based on a GPU-accelerated parallel symplectic method was developed to achieve rapid and accurate defects location after identification.Finally,the accuracy and reli-ability of the technology were verified through numerical simulation and engineering application.The results show that the average identification accuracy of this technology for various underground hidden defects ex-ceeds 88%,and the location error is controlled within 5%.This technology provides robust technical support for the rapid and accurate detection and diagnosis of underground hidden defects in urban road.关键词
路基隐蔽病害/识别定位/探地雷达/YOLOv11-CL/RTMKey words
underground hidden defects/identification and location/ground penetrating radar(GPR)/YO-LOv11-CL/reverse time migration(RTM)分类
交通工程引用本文复制引用
方宏远,黄凯,万静茹,雷建伟,胡浩帮,王念念,薛冰寒..基于YOLOv11-CL和GPU并行RTM的城市道路路基隐蔽病害快速识别定位[J].东南大学学报（自然科学版）,2026,56(10):1392-1403,12.基金项目
国家自然科学基金资助项目(52408406) （52408406）
国家重点研发计划资助项目(2022YFC3801000) （2022YFC3801000）
中国博士后科学基金资助项目(2024T170841) （2024T170841）
河南省科技研发计划联合基金重点资助项目(235200810014). （235200810014）
