Tunnel Deformation Monitoring System Based On Vision Assistant

Support by 973 Subject No.2013CB035405,this dissertation undertakes tunnel deformation monitoring of the TBM operations in "the principle of sensing state in the hard rock drivage".With the development of China,the demand for TBM(Tunnel Boring Machine)in urban traffic,hydraulic engineering and municipal construction grows gradually.China has become the largest market for TBM,accounting for 50%of global usage.The construction safety is one of the main consideration of the TBM operations in which the deformation of the tunnel is the main factor affecting the construction security.Excessive deformation of the tunnel will cause tunnel landslide and the jamming of TBM.So it's important to use tunnel deformation monitoring system in the TBM operations.A tunnel deformation monitoring system based on vision assistant is proposed to automati-cally,accurately and timely monitoring the deformation of tunnel.Combined with the condition of TBM construction and the goal of monitoring,the method of aiming at targets by using fluores-cent paint to mark,a camera to detect and a Pan-Tilt to drive is developed.A stereo vision system improving the efficiency of measuring and a method of stitching three-dimensional point clouds are developed.Firstly,the general situation of the TBM and the problem brought by the tunnel deformation in the TBM operations is introduced.The principles as well as insufficiencies of the domestic and international tunnel deformation monitoring methods by different classification are introduced.The background and significance of research is proposed.Then,the design of artificial mark and experiment device are presented.The kinematics model of device and the vision model of camera are built.Image processing methods,such as image de-noising,threshold segmentation,erosion and dilation,are used to obtain the centroid coordinates of targets.An image-based visual servo control is proposed to aiming at targets with the laser displacement sensor.The stereo vision system is used to measure the relative distance between points and the laser displacement is used to measure sensor measure three-dimensional coordinates of some points to increase efficiency of measuring with semi-definite programming.The influences of parameters on accuracy are analysed and appropriate parameters are chosen to actual measurement.A method of stitching three-dimensional point clouds at different times are preliminary studied.Finally,The feasibility of the measurement system is verified through simulation experiment.