摘要
裂缝、渗水和掉块等隧道病害对隧道运行安全和使用年限有着严重影响,传统的隧道裂缝检测方法以人工检测为主,速度慢、漏检严重且无法对隧道裂缝进行长期自动化检测。本文设计了一套基于深度神经网络的面向隧道裂缝的检测系统,该系统使用隧道面阵相机采集的图像数据对隧道裂缝进行检测识别。该检测系统能够对隧道裂缝快速准确地进行检测,在检测阈值设置为0.3时,隧道裂缝的检测准确率为97.00%。经工程验证,本文所设计的系统对隧道主要病害裂缝病检测精度达到0.2 mm,满足隧道检测任务需求。
Tunnel diseases such as cracks,water seepage and falling blocks have a serious impact on the operation safety and service life of the tunnel.The traditional tunnel disease detection methods are mainly manual detection,with slow speed and serious missed detection,and can not detect the tunnel diseases automatically for a long time.This paper designed a tunnel disease detection system based on area scan camera and joint detection algorithm,which could quickly detect tunnel diseases and explored database technology to monitor the detected diseases.After engineering verification,the detection accuracy of the system for the main diseases and cracks of the tunnel was 0.2 mm,which met the requirements of tunnel detection tasks.
作者
尹剑
陈树翔
王洪战
YIN Jian;CHEN Shuxiang;WANG Hongzhan(Anhui Guoju Construction Machinery Technology Company Limited,Hefei Anhui 230000,China;China Railway Qinghai-Tibet Group Company Limited,Xining Qinghai 810000,China;China Railway Liu Yuan Group Company Limited,Tianjin 300308,China)
出处
《北京测绘》
2023年第7期1032-1036,共5页
Beijing Surveying and Mapping
基金
青藏集团科研开发计划(QZ2021-G01)
中铁六院重点课题(KY-2021-15)。
关键词
隧道病害检测
数据采集
神经网络
目标检测
语义分割
tunnel disease detection
data acquisition
neural network
object detection
semantic segmentation