A digital twin approach for tunnel deformation detection
To solve the problem of slow tunnel deformation and difficulty in obtaining effective experimental detection data leading to limited research on tunnel deformation detection technology,a digital twin method for tunnel deforma-tion detection is proposed,and a high-fidelity tunnel twin model is establishedin this paper.The deformation of tunnel is simulated by finite element method,and the true value of tunnel twin model is obtained.A virtual simulation plat-form is built to realize the three-dimensional laser scanning of tunnel models in a virtual environment to obtain large sample detection data and assist in training deformation detection methods.In deformation detection,Geotransformer neural network is used to realize tunnel point cloud registration,and tunnel section point cloud is obtained by fitting tunnel central axis to realize tunnel deformation analysis.Experimental results show that the proposed method can ef-fectively overcome the problem of tunnel deformation detection technology research limited by experimental sites.The average error of tunnel model surface reconstruction is 0.00253 mm and the maximum error is 1.1325 mm.Compared with the true value of deformation output by finite element method,the average error of deformation detection is less than 0.34 cm.It is verified that the deformation detection method has high accuracy and basically meets the engineer-ing requirements.
measurementdigital twintunnel deformation detectionsurface reconstructionfinite elementlaser point cloud
万方数据
维普
