基于计算机视觉的盾构隧道管片错台自动测量方法
作者简介:
吴庆杰,工程师,E-mail: wuqingjiew@ 163. com
作者单位:
1.福州地铁集团有限公司,福建福州 350004; 2.中交第二航务工程局有限公司,湖北武汉 430040;3.同济大学土木工程学院地下建筑与工程系,上海 200092
基金项目:
∗国家自然科学基金(52038008)
摘要:
盾构隧道由预制管片拼装而成,但接头拼装误差会引发管片错台,若达到承载力极限状态将影响结构正常使用。长距离盾构隧道结构存在大量管片接头,目前缺乏有效手段快速测量工程全域。因此,基于三维激光扫描点云,采用计算机视觉技术,提出盾构隧道管片错台自动测量方法。首先,通过三维深度学习技术实现盾构隧道点云智能分割,分割总精度达 94.0%.随后,对各管片开展几何性状拟合,实现自动计算管片间的错台量,并应用于实际工程中,通过与现场人工测量结果进行对比,验证所提方法的准确性,实现高精度错台自动测量。

English:
The shield tunnels are assembled by prefabricated segments, joint assembly erros can causethe segmental dislocations if it reaches the ultimate limit state of bearing capacity, it will affect the normaluse. There are a large number of joints for shield tunnels, it is still a lack of efficient means to measurethe dislocations of numerous joints. Therefore, based on the 3D laser scanning point colud, computervision technology is used, an automatic measurement method of segmental dislocations is proposed.Firstly, the segmentation of tunnel point clouds is realized by 3D deep learning technology,with anoverall accuracy of 94.0%. Subsequently, the geometry of each segment is fitted to calculate thedislocation between the segments. It is applied in actual engineering, by comparing with the on-sitemanual measurement results, the accuracy of the proposed method is verified, achieving high-precisionautomatic measurement of the dislocation.