CN-116295081-B - Tunnel total space deformation extraction method based on mobile laser scanning

CN116295081BCN 116295081 BCN116295081 BCN 116295081BCN-116295081-B

Abstract

The invention discloses a tunnel full-space deformation extraction method based on mobile laser scanning, which comprises the steps of detecting a section deformation value S_ D of a tunnel at each sampling position by using a tunnel section instrument, recording a tunnel section position L_ D and a section measuring line angle value G_ D , scanning the whole tunnel space by using a vehicle-mounted mobile laser scanner to obtain a section deformation value S_ Y of each position of the tunnel space, taking the section deformation value S_ D obtained by the tunnel section instrument at the sampling position as output, taking the section deformation value S_ Y obtained by scanning by the mobile laser scanner at the same position as input, constructing a sample set, training a XGBoost machine learning model by using a sample set, inputting the section deformation value S_ Y at each position of the tunnel space into the trained XGBoost machine learning model after training is completed, and taking the output deformation value S_ D as the section deformation value of the corresponding position. The tunnel profiler data and the global movement three-dimensional laser scanning data are fused, a correlation mechanism of deformation of a sampling position and global deformation of a tunnel is revealed, and the rapid extraction of the total spatial deformation of the tunnel is realized.

Inventors

WANG QIANG
YUAN DONGYANG
TONG JUNHAO
ZHANG WEIKANG
CHEN AIQING
ZHAI JUNLI
WANG HAOZHENG

Assignees

浙江省交通运输科学研究院

Dates

Publication Date: 20260512
Application Date: 20230406

Claims (8)

1. The tunnel full-space deformation extraction method based on mobile laser scanning is characterized by comprising the following steps of: s1, detecting section deformation values of a tunnel at each sampling position by using a tunnel section instrument And record the tunnel section position Angle value of section survey line ; S2, scanning the whole tunnel space by using a vehicle-mounted mobile laser scanner of the tunnel detection vehicle, and obtaining section deformation values of all positions of the tunnel space ; S3, using section deformation values obtained by a tunnel section instrument at a sampling position For outputting, the section deformation values obtained by the scanning of the laser scanner are moved at the same position For input, constructing a sample set; S4, training XGBoost the machine learning model through sample set, and after training, deforming the section of each position in the tunnel space Inputting the trained XGBoost machine learning model and outputting deformation values As a section deformation value of the corresponding position; section deformation value The acquisition process of (a) is specifically as follows: preprocessing the laser point cloud obtained by scanning the vehicle-mounted mobile laser scanner, Acquiring the position of a cross section of a vehicle-mounted mobile laser scanner Within a set distance between the front and rear of the cross section measuring line angle value Sector point cloud data in the vicinity; Projecting the sector point cloud data along the longitudinal direction of the tunnel, taking the set distance as a calculation unit, acquiring inner and outer boundary point clouds formed after the projection of the sector point clouds, fitting the inner and outer boundary point clouds to obtain inner and outer boundary contour fitting lines, and calculating a section deformation value which is the section deformation value obtained by calculating the average value of the inner and outer boundary contour fitting line coordinates 。
2. The method for extracting the tunnel total space deformation based on the mobile laser scanning as claimed in claim 1, wherein the vehicle-mounted mobile laser scanner is arranged at the section position Within a set distance between the front and rear of the cross section measuring line angle value Averaging the nearby deformation data to obtain the section position Cross section deformation value of (2) 。
3. The method for extracting the full-space deformation of the tunnel based on the mobile laser scanning as claimed in claim 1, wherein the preprocessing comprises removing the discrete point cloud and the interior object point cloud in the tunnel.
4. The method for extracting the tunnel total space deformation based on the mobile laser scanning as claimed in claim 1, wherein the sampling position acquisition method is specifically as follows: Dividing the tunnel into a straight line tunnel and an arc tunnel with different curvature radiuses, and setting the intersection point of the straight line tunnel and the arc tunnel and the intersection point of the arc tunnels with different curvature radiuses as sampling positions; aiming at the linear tunnel, sampling positions are taken at intervals of a set distance; for an arc tunnel, a sampling interval is set based on the curvature radius of the arc tunnel, and sampling positions are taken based on the sampling interval.
5. The method for extracting full-space deformation of tunnel based on mobile laser scanning as claimed in claim 1, wherein if the tunnel cross section is a single slope, the value measured by the tunnel profiler is the tunnel cross section deformation value If the tunnel cross section is a double slope, calculating the deformation value of the tunnel cross section by using a trigonometric function according to the geometrical position relationship between the slope surface gradient and the height of the tunnel section instrument 。
6. The method for extracting full-space deformation of tunnel based on mobile laser scanning as claimed in claim 1, wherein the position of the section of the tunnel is identified by a mileage stake mark of the tunnel Position of tunnel section 。
7. The method for extracting the full-space deformation of the tunnel based on the mobile laser scanning as claimed in claim 1, wherein the tunnel is a mountain tunnel.
8. The tunnel full-space deformation extraction method based on mobile laser scanning according to claim 1, wherein a sample set is randomly divided into 5 parts, the number n_tree of super-parameter decision trees and the depth n_depth of decision trees of XGBoost machine learning models are set to take value ranges, 4 parts of data are selected as model training sets, model training is carried out, the rest 1 part is used as a verification set to carry out prediction accuracy verification, the method is repeated 5 times, the prediction error of the models is evaluated through root mean square error, the smaller the root mean square error of the test set is, the higher the model accuracy is, and the models with high accuracy are selected as the trained XGBoost machine learning models.

Description

Tunnel total space deformation extraction method based on mobile laser scanning Technical Field The invention belongs to the technical field of tunnel monitoring, and particularly relates to a tunnel total space deformation extraction method based on mobile laser scanning. Background Along with the continuous promotion of the strategy of 'traffic strong countries', the mileage and the number of tunnel traffic in China are rapidly increased, people must be consciously aware that a plurality of tunnels built in early years are exposed to more and more problems while feeling the tired fruits, the development of the tunnels is changed from the construction to the construction and the construction, so that the analysis and the evaluation of the health state of the tunnel structure are more and more important, and the deformation of the section of the tunnel is an important parameter for realizing the analysis and the evaluation of the health state of the tunnel structure. At present, the mountain tunnel convergence deformation is mostly dependent on the traditional detection method, namely, a technician detects the deformation of the position of the monitoring point through a tunnel section instrument, and the traditional detection method has the following problems: (1) The full-space deformation of the tunnel cannot be timely and comprehensively acquired; (2) The manual monitoring means needs to consume a great deal of manpower and material resources and has low efficiency. Disclosure of Invention The invention provides a tunnel total space deformation extraction method based on mobile laser scanning, and aims to solve the problems. The invention is realized in such a way that a tunnel full-space deformation extraction method based on mobile laser scanning comprises the following steps: S1, detecting a section deformation value S_ D of a tunnel at each sampling position by using a tunnel section instrument, and recording a tunnel section position L_ D and a section measuring line angle value G_ D; S2, scanning the whole tunnel space by using a vehicle-mounted mobile laser scanner of a tunnel detection vehicle, and obtaining section deformation values S_ Y of all positions of the tunnel space; S3, taking a section deformation value S_ D obtained by a tunnel section instrument at a sampling position as output, and taking a section deformation value S_ Y obtained by scanning a mobile laser scanner at the same position as input to construct a sample set; S4, training XGBoost the machine learning model through a sample set, inputting the section deformation value S_ Y of each position of the tunnel space into the trained XGBoost machine learning model after training is completed, and taking the output deformation value S_ D as the section deformation value of the corresponding position. Further, the deformation data of the vehicle-mounted mobile laser scanner, which is located near the section line angle value g_ D within the set distance between the front and rear of the section position l_ Y, is averaged to obtain the section deformation value s_ Y of the section position l_ Y. Further, the process of obtaining the section deformation value s_ Y is specifically as follows: preprocessing the laser point cloud obtained by scanning the vehicle-mounted mobile laser scanner, Acquiring sector point cloud data of the vehicle-mounted mobile laser scanner, which are positioned near a section measuring line angle value G_ D within a set distance before and after a section position L_ Y; And projecting the sector point cloud data along the longitudinal direction of the tunnel, taking the set distance as a calculation unit, acquiring inner and outer boundary point clouds formed after the projection of the sector point clouds, fitting the inner and outer boundary point clouds to obtain an inner and outer boundary contour fitting line, and calculating a section deformation value which is the section deformation value S_ Y by the average value of the inner and outer boundary contour fitting line coordinates. Further, the preprocessing comprises removing the discrete point cloud and the interior decoration point cloud in the tunnel. Further, the sampling position acquisition method specifically comprises the following steps: Dividing the tunnel into a straight line tunnel and an arc tunnel with different curvature radiuses, and setting the intersection point of the straight line tunnel and the arc tunnel and the intersection point of the arc tunnels with different curvature radiuses as sampling positions; aiming at the linear tunnel, sampling positions are taken at intervals of a set distance; for an arc tunnel, a sampling interval is set based on the curvature radius of the arc tunnel, and sampling positions are taken based on the sampling interval. Further, if the tunnel cross section is a single slope, the value measured by the tunnel section instrument is the tunnel section deformation value S_ D, and if the tunnel cross sect