基于探地雷达的路基无损检测技术研究
Research on Non-Destructive Testing Technology for Roadbed Based on Ground Penetrating Radar

随着城市地下空间资源的开发力度日益增大，地层不均匀沉降等城市道路灾害频发，路面坍塌的主要原因是路基中存在空洞病害，因此对地下空洞的有效探测尤为重要。探地雷达是目前路基无损检测的一种较为成熟的装置，但采集得到的探地雷达图像并非直接成像，很大程度上依赖数据处理人员的经验水平，使得数据解释耗时且容易误差增大。本文针对路基中空洞的检测问题，对二三维探地雷达分别进行阐述和对比，在三维探地雷达的基础之上提出了探地雷达和多种方法联合的无损检测技术。该技术可通过深度学习、机器学习和注意力融合等多种方法对探地雷达采集到的图像数据进行处理，实现对路基空洞的准确识别，从而形成精确、可靠、高效的路基健康检测技术。
With the increasing development of urban underground space resources and the frequent occurrence of urban road disasters such as uneven settlement of the ground, the main reason for road collapse is the existence of cavity diseases in the roadbed, so the effective detection of underground cavities is particularly important. Ground penetrating radar (GPR) is a relatively mature device for nondestructive testing of roadbed at present. However, the collected GPR images are not directly imaged and largely depend on the experience level of data processing personnel, making data interpretation time-consuming and prone to increased errors. In this paper, aiming at the detection of voids in roadbed, two and three dimensional ground penetrating radars are described and compared, and a non-destructive detection technology based on three-dimensional ground penetrating radars and multiple methods is proposed. This technology can process image data collected by ground penetrating radar through multiple methods such as deep learning, machine learning, and attention fusion to achieve accurate recognition of roadbed voids, thereby forming an accurate, reliable, and efficient roadbed health detection technology.