Extraction Of Coal Mining Subsidence Information Based On Airborne Laser Scanning Feature Point Cloud

Underground coal mining causes subsidence of the surface above the working face,causing problems such as ecological environment,damage to facilities,and natural disasters.Traditional mining subsidence monitoring methods are inefficient and can only obtain linear subsidence data.Airborne LiDAR technology has significant advantages in obtaining-mining subsidence surface information,but in practical applications there are problems such as point cloud filtering and significant errors in subsidence modeling.To this end,this paper takes the surface of two fully mechanized mining faces in Yushen mining area as the experimental area,uses the airborne LiDAR system to collect multi-period point cloud data,and obtains the ground feature point cloud by improving the point cloud filtering algorithm and deep learning method to construct high-precision subsidence model and extract horizontal movement information.The main content and results of the thesis research are as follows:(1)An improved algorithm for point cloud filtering is proposed in combination with terrain features,which improves the effect of ground point cloud filtering.Common point cloud filtering algorithms are selected for comparison,and the results show that the use of triangulation progressive encryption filtering algorithm can better eliminate vegetation and other non-ground points and retain ground points.The optimization of the two parameters is improved on the basis of the triangulation progressive encryption filtering algorithm.The results show that the improved algorithm reduces the error and is more suitable for the vegetation coverage conditions of the Yushen mining area.(2)The use of terrain factors combined with DNN models to extract feature point clouds reduces the error impact of complex landforms on subsidence modeling.By analyzing topographic factors and constructing DNN models,the feature areas that are less affected by topography are extracted,and the DEM of the complete subsidence basin is obtained by interpolation.The GaussAmp function is used to model the surface of the subsidence basin.The results show that the accuracy of using local polynomial interpolation algorithm in the non-sinking area and the radial basis function interpolation algorithm in the subsidence area to construct the subsidence basin is higher;the GaussAmp function based on the LM algorithm has the highest accuracy in fitting the basin.(3)By improving the existing feature matching algorithm,the technical approach of extracting horizontal movement information based on the subsidence model is discussed.The feature matching algorithm is improved based on the BSC(binary shape context)operator combined with terrain factors.Analyze the influence of point cloud density and parameters on the algorithm,extract the horizontal movement of the surface in the coal mining subsidence area,and analyze the quantitative relationship between the horizontal movement deviation and the point cloud density and topographic characteristics.The results show that the horizontal movement curve extracted by the improved feature matching algorithm conforms to the basic law of the horizontal movement of coal mining subsidence,and the terrain factor with strong correlation with the horizontal movement deviation can be used to measure the magnitude of the horizontal movement extraction error of the improved feature matching algorithm.The research results of the thesis provide technical support for airborne laser scanning to obtain accurate three-dimensional deformation information of coal mining subsidence area.