Concrete Structure Crack Localization Tracking And Automatic Finite Element Modeling Based On Point Clouds

Structural analysis and assessment is an important part of civil engineering structure maintenance,which can effectively judge the health status of structures and find structural defects and hidden dangers.Based on the three-dimensional information advantages of point cloud model data of civil engineering concrete structures,this study integrated the detection,analysis,and maintenance requirements of concrete structures during service in structural intelligent analysis and assessment.Moreover,this study introduced and proposed systematic point cloud data processing algorithms and technical means.It is committed to exploring the research of concrete structure point cloud data in crack localization tracking and automatic finite element modeling and analysis,covering the following main contents:(1)Focusing on the characteristics of large volume,unobvious surface features,and similar components of concrete structures,a monocular 3D reconstruction optimization framework for concrete structures was proposed,which realized the generation of high-quality point cloud models of target structures.In addition,the lightweight of model data was realized by combining the geometric down-sampling algorithm,which reduced the data volume of the point cloud model while retaining its data shape features.(2)A viewpoint localization algorithm and a mapping registration algorithm were proposed to realize the crack localization,quantization,and tracking based on point cloud data directly,and a Web front-end virtual space integrating algorithms and point cloud data was established.Taking a concrete structure as an example,this method took its point cloud model as the world reference system,calculated the world coordinates of cracks through viewpoint localization,and realized the world coordinates of visual index cracks based on crack images.Moreover,the crack point cloud model was mapped and registered for the point cloud model of the concrete structure to realize the width quantification and long-term tracking of cracks during the expansion.The test results demonstrated that the average localization time of the crack was 38.09 μs,and the width quantization error was less than 8%.(3)A modeling parameter set extraction method based on Point Net++ point cloud segmentation network and multiple point cloud data feature extraction algorithms was proposed.A three-dimensional modeling data reconstruction method for the inner structure was established,and the automatic transformation of the point clouds of the concrete structure to the three-dimensional solid model was realized.Taking the point clouds of in-service concrete bridge structure as an example,this method proposed slice projection,ordered concave package,and other algorithms to obtain the cross-section feature data of components,and introduced a parametric modeling method to realize the automatic modeling of structure.The test results demonstrated that the size average deviation of piers,cap beams,bridge decks,and the inner structure is 0.46%,0.42%,0.15%,and 0.50% respectively.(4)Based on the solid model of concrete bridge generated by the point cloud model,the automatic transformation and analysis method of the finite element model(Scan to FEM)was further established.The method was based on the solid model generated in this study,which was divided by hexahedral mesh,then the boundary conditions and multi-stage loading vehicle load conditions were set,and the finite element analysis of the model was carried out.Finally,the load-deflection curve,crack development,and material stress in each state obtained from the test demonstrated that it is feasible to analyze and assess the structural stress state based on the point clouds of civil engineering structures.