Quality Evaluation Of Rock Mass And Finite Element Modeling In Dagushan Open-pit Mine Based On Tilt Photogrammetry

As a widely used method to solve the stability of geotechnical engineering,the accuracy of analysis model and the efficiency of modeling directly affect the result of numerical simulation.The establishment of the analysis model includes the establishment of the grid model,the determination of the mechanical parameters of rock mass,the setting of boundary conditions and initial conditions.The efficiency of modeling large scale models can be improved by using tilt photogrammetry technology to build a numerical model based on a three-dimensional real model.The efficiency of finite element modeling can be improved by establishing finite element model based on 3D real scene model establisedd by tilt photography technology.However,the tilt photography model is a triangular plate model with vegetation,equipment and other disturbances,while the numerical analysis model is a regular grid model.So the tilt photography model cannot be directly converted to a grid model.On the other hand,the artificial classification method of rock mass quality cannot meet the production demand of large-scale open pit mines,and the automatic classification method of rock mass quality needs to be proposed.In order to solve the above problems,this paper proposes a grid model establishment method based on the tilt photography model and a classification method based on semantic segmentation,so as to realize the establishment of a fine numerical calculation model based on the tilt photography technology in Dagushan open-pit mine.The main work is as follows:(1)Tilt photogrammetry in Dagushan open-pit mine.UAVs were used to carry five lenses for tilt photogrammetry and 3D real scene modeling of Dagushan open-pit mine.(2)Due to the large amount of model data and redundancy,it can’t be directly used in numerical calculation.To build a grid model,direct filtering methord is used to obtain the contour point cloud,the digital iamges processing technology is used to regularize the contour line,and then the entity model is reconstructed and meshed with Rhino software.(3)Rock mass structure discrimination of Dagushan open-pit mine based on semantic segmentation networkAccording to the classification standard of rock mass structure proposed by Gudzhen,the rock mass structure types can be divided into four categories:Integrated or block structure,layed structure,breaking structure,granular structure.The image collected by UAV is used to make data set,and the semantic segmentation network is trained to realize automatic discrimination of rock mass structure type.Applying the trained semantic segmentation network to the gradient photography model with clear texture can realize the automatic recognition of the rock mass structure of the open pit slope.(4)GSI block model establishment.The rock mass structure type of Gudezhen corresponds to the rock mass structure type of rock mass strength index GSI.And the GSI value of rock mass is determined by determining the structural plane conditions based on the geological survey data.For the GSI value of the unknown area which is covered by scum and vegetation,The GSI value is determined by the GSI value of the known area by the interpolation method.By this method,the GSI block model of Dagushan slope is established...