Subway Tunnel Deformation Monitoring Based On3D Laser Scanning Data

In recent years, subway has been the main mean of public transportin large and medium-sized cities because of the advantages of highefficiency, energy-saving and steady. But the subway lines alwaysdistributing in the urban center, the security problem might become thepotential safety hazard if we neglect it. So a large technical demand isexisting in the field of real-time monitoring of subway tunnel; On theother hand, in terms of long operating time and high departure frequency,there is a higher requirement for the model and efficiency of monitoringsystem. However, intelligent total station is the general approach fordeformation monitoring of subway tunnel in current use, but the data sizeis very limited during information extracting, a method for deformationmonitoring based on3D point cloud is proposed in this paper.In this paper, data source came from the terrestrial point clouds.The method include four parts: point cloud registration, central axisdetermining, vertical section abstraction and deformation curve fitting.Considering the structure characteristic of tunnel is long and narrow,the traditional adjacent stations registration is replaced bysection-controlled registration, in this way the common control pointscan be shared for each station and the errors accumulation can be avoided;The central axis of subway tunnel can be determined through RANSAC (RandomSample Consensus) algorithm and least square adjustment; The orientationof tunnel firstly has to be adjusted in terms of the central axis, thenthe data can be fitted by local quadric fitting, which include constrainedleast squares algorithm, RANSAC algorithm and Robust Estimation. At last,the deformation curve can be fitted to the deformation values throughcubic spline; The radius of fitting circle for vertical section can becalculated and used for the comparison of adjacent circles.The feasibility and accuracy of the proposed method have been verifiedusing the point clouds acquired by RIEGL VZ-400laser scanner. Theexperiment is consisted of section control registration, fitting analysisand abstraction efficiency. A section-controlled registration experimenthas been made in a110meters long corridor, and the error accumulation can be eliminated; The fitting analysis experiment makes calculation ofinterpolated accuracy and radius comparison of adjacent vertical sections,the maximum error between interpolated point and real point is1.5mm, andthe minimum one is0.1mm; The convergent tendency of tunnel can be detectedby the comparison of adjacent fitting radius, the maximum error is6mm,the minimum one is1mm; Measuring time for continuous vertical sectionabstraction during the abstraction efficiency experiment, eachabstraction cost3seconds at the most. It is proven by experiments thatthe accuracy requirement of point cloud registration and vertical sectionabstraction can be satisfied by the method proposed in this paper, as wellthe deformation can be detected precisely and rapidly; Furthermore,compared with the traditional approaches, the3D laser scanningtechnology has a obvious advantage in respect of data size, dataacquisition rate, human cost, etc.