基于无人机技术的大坝表面裂缝识别技术研究  

Research on dam surface crack identification technology based on UAV technology

作　　者：张振[1] 徐梓源 鲍永龙 张少中 Zhang Zhen;Xu Ziyuan;Bao Yonglong;Zhang Shaozhong(Engineering Construction Branch of Upper Yellow River Hydropower Development Co.,Ltd.,813300,China;College of Hydraulic and Electric Engineering,Qinghai University,810016,China;China Hydropower Fourth Bureau(Lanzhou)Machinery and Equipment Co.,Ltd.,731600,China)

机构地区：[1]黄河上游水电开发有限责任公司工程建设分公司,西宁813300 [2]青海大学水利电力学院,西宁810016 [3]中国水电四局(兰州)机械装备有限公司,兰州1731600

出　　处：《青海交通科技》2023年第5期66-72,103,共8页Qinghai Transportation Science and Technology

摘　　要：水工混凝土结构建筑物,如混凝土坝,在使用过程中会因交替荷载和材料老化而受到损坏形成表面裂缝。裂缝的产生可能会给大坝的安全性带来挑战。因此,监测水工混凝土结构裂缝的变化状态对水利工程的健康服务具有重要意义。研究结合无人机技术和数字图像处理的方法,提出了一种基于水工结构的自动损伤检测和诊断方法。它被设计用于表面损伤的检测和分割,具体来说,为了提高裂纹特征提取效果,基于YOLOv5使用了ResNet101作为主干网络,对裂缝进行特征提取。与静态图像和直接人工测量相比,无人机的机载图像具有更高的精度,这表明其在实现远程、可靠和快速的裂纹检测方面具有广泛的应用潜力。Hydraulic concrete structures,such as concrete dams,will be damaged by alternating loads and aging materials in the process of use to form surface cracks.The generation of cracks may bring challenges to the safety of DAMS.Therefore,monitoring the changing state of hydraulic concrete structure cracks is of great significance to the health services of hydraulic engineering In this study,an automatic damage detection and diagnosis method based on hydraulic structure was proposed by combining UAV technology and digital image processing method.It was designed for surface damage detection and segmentation.Specifically,in order to improve the crack feature extraction effect,ResNet101 was used as the backbone network based on YOLOv5 for crack feature extraction Compared with static images and direct manual measurements,airborne images of unmanned aerial vehicles(UAVs)have higher accuracy,which indicates that it has wide application potential in realizing remote reliable and rapid crack detection.

关 键 词：无人机 数字图像处理 裂纹特征 无损检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TV698.1[自动化与计算机技术—计算机科学与技术] V19[水利工程—水利水电工程]