基于机器视觉的盾尾间隙智能检测方法研究  

Intelligent detection method of shield tail clearance based on machine vision

作　　者：胡秋斌 毛仁利 庄欠伟 柳献[1] HU Qiubin;MAO Renli;ZHUANG Qianwei;LIU Xian(School of Civil Engineering,Tongji University,Shanghai 200092,China;Shanghai Tunnel Engineering Co.,Ltd.,Shanghai 200232,China)

机构地区：[1]同济大学土木工程学院,上海200092 [2]上海隧道工程有限公司,上海200232

出　　处：《铁道科学与工程学报》2025年第3期1357-1368,共12页Journal of Railway Science and Engineering

基　　金：上海市国资委企业创新发展和能级提升项目(2022020);同济大学学科交叉联合攻关项目(2023-2-ZD-03)。

摘　　要：盾尾间隙为盾构掘进姿态调整、管片选型等提供重要依据,为了解决盾构施工中盾尾间隙人工测量难度高、准确度难以保证的问题,提出一套盾尾间隙智能检测方法,可以实现施工中盾尾间隙的实时测量。该方法采用安装在拼装机回转机构上的视觉扫描系统获取管片和盾尾的点云数据。对点云数据进行滤波降噪后提取管片手孔部分点云,以此确定盾尾间隙检测位置。采用PSO算法对管片内弧面点云进行柱面拟合,精确求解管片轴线位置。利用管片轴线和盾尾间隙检测位置点确定的径向平面提取管片点云和盾尾点云的切片,采用改进的最小二乘法求解所得切片上两点云的直线方程。计算两直线在管片端面处的空间距离即为盾构间隙与管片厚度之和,继而可得到检测位置的盾构间隙。对盾尾间隙计算方法进行研究,结果表明:相对于非线性最小二乘拟合方法,PSO算法在对管片点云拟合圆柱模型中精度更高,求解管片柱面的轴线位置更准确;采用RANSAC算法对传统最小二乘法进行改进,可以提升拟合盾尾间隙所在平面与管片或盾尾交线的准确度,有利于提升盾尾间隙解算精度;经现场测试验证,本文的盾尾间隙测量系统重复性测量精度优于1.3mm,绝对测量精度优于2.4mm,可满足工程测量要求。研究可为盾构隧道施工中的盾尾间隙提供一种新的自动化测量技术,促进盾构隧道施工自动化、智能化。The shield tail clearance provides crucial data for adjustments in shield tunneling posture and selection of segment assembly point.Manual measurement of shield tail clearance during shield construction is often challenging and imprecise.To address these issues,this article proposed an intelligent detection method that enables real-time measurement of shield tail clearance during construction.A visual scanning system mounted on the rotary mechanism of the assembly machine captures point cloud data of the segments and shield tail.Following the filtering and denoising of the point cloud data,the point cloud of the segment's handhole area was extracted to determine the inspection position for the shield tail clearance.The PSO algorithm was employed to fit the inner arc surface point cloud of the segment to a cylindrical surface,thus accurately determining the segment's axis position.A radial plane,defined by the segment's axis and the shield tail clearance inspection point,was used to extract slices of the segment and shield tail point clouds.An improved least squares method was applied to solve the linear equations of the two point clouds on the obtained slices.The spatial distance between the two lines at the end face of the segment,which was the sum of the shield clearance and the segment thickness,was then calculated.It could determine the shield clearance at the inspection position.This article also explored methods for calculating the shield tail clearance,demonstrating that the PSO algorithm achieves higher accuracy in fitting the cylindrical model to the segment point clouds compared to the non-linear least squares fitting method,thereby more accurately determining the position of the segment axis.By employing the RANSAC algorithm to enhance the traditional least squares method,the accuracy of fitting the intersection line between the plane where the tail clearance was located and the segment or the shield tail was improved,which enhanced the precision of solving the shield tail clearance.Field tests have ve

关 键 词：盾尾间隙 智能检测 机器视觉 点云 PSO算法 RANSAC算法 

分 类 号：U45[建筑科学—桥梁与隧道工程] P231[交通运输工程—道路与铁道工程]