Research On Ground Subsidence Monitoring Method Of UAV Photogrammetry And LiDAR In Mining Area

Coal is one of the main energy sources to ensure China’s social and economic development.Long-term and high-intensity mining leads to increasingly serious surface subsidence and environmental problems.The movement and destruction of overlying rock mass caused by mining is a complex process in time and space.Therefore,strengthening the real-time dynamic monitoring of surface subsidence in the mining area is conducive to timely grasp the law of surface subsidence in coal mining,which is of great significance for the protection of ecological environment and intelligent mining of coal mine.The development of measurement technology promotes the acquisition of high-resolution terrain data.UAV photogrammetry is flexible and efficient with low cost.Airborne LiDAR can monitor all day and high-precision.Therefore,in view of the problems of low efficiency,high labor intensity and unable to form the surface subsidence basin in the whole working face affected area of the traditional observation technology in the subsidence monitoring of mining area,in this paper,the UAV equipped with visible light camera and LiDAR sensor is used to conduct multi-period comprehensive monitoring of the surface basin formed by coal mining subsidence.The problem of obtaining high-precision DEM is further refined and solved,which provides a technical innovation way for ground monitoring of coal mining subsidence.The main research contents and results are:(1)This paper summarizes the research status of UAV subsidence monitoring,UAV georeferencing method and point cloud filtering algorithm at home and abroad,points out the shortcomings in these aspects,expounds the key technologies in UAV subsidence monitoring,summarizes the data acquisition and general processing flow of UAV photogrammetry and airborne LiDAR,and compares and analyzes the differences between the two technologies.(2)Aiming at the influence of the distribution density of ground control points(GCP)on the accuracy of UAV photogrammetry results,this paper carried out the photogrammetry experiment based on FEIMA D2000 UAV,and proposed the accuracy evaluation method and workflow of the results.The position accuracy of DSM and DOM obtained by direct georeferencing reference method without ground control and the relationship with the number and distribution of GCP were studied.The results show that when the GSD is 1.7cm/pixel,the vertical RMSE of DSM obtained by UAV direct georeferencing method is 0.087 m,which is about 5GSD,and the horizontal RMSE of DOM is 0.041 m,which is about 2.4GSD.The UAV direct georeferencing method reduces the dependence on the ground layout control points in the bundle adjustment.It is a flexible and efficient method to obtain high-resolution terrain data.(3)Aiming at the filtering problem of removing non-ground points from UAV dense matching point cloud,this paper evaluates the performance of five different ground point filtering algorithms implemented in widely used software from both qualitative and quantitative aspects,and studies the influence of point cloud density on the filtering results.The results show that the success of ground point cloud filtering algorithm depends on the selection of optimal parameters to a certain extent.The five filtering methods can remove most of the non-ground points,and no method can perform best in both retaining the ground points and removing the non-ground points.Among them,the adaptive triangulated irregular network(ATIN)algorithm obtains the minimum total error in both test areas by better removing non-ground points.The influence of point density on the filtering results is relatively small.Generally,when the resolution of data is reduced,the separation of ground points and objects will be more difficult.In the absence of airborne LiDAR data,dense matching point cloud is also a good alternative.(4)Aiming at the problem that the traditional observation technology cannot form a surface subsidence basin,airborne LiDAR point clouds are used as reference data for collaborative registration of UAV multi-temporal dense matching point clouds based on their penetrability in vegetation areas.The performance of the co-registration is evaluated by calculating the M3C2 distance.Then,based on the differential DEM obtained by the filtering and interpolation of the UAV dense matching point cloud,a dynamic subsidence basin for surface subsidence monitoring is constructed,and the development process and subsidence law of the moving basin are studied.The accuracy of the dynamic subsidence basin is analyzed by comparing with the measured points on the ground.The uncertainty of differential DEM is quantified.The results show that the RMSE between the elevation difference of the monitoring points on the four subsidence DEMs and the subsidence difference of the measured data is mostly between 0.2 m and 0.3 m,and the highest accuracy can reach 0.17 m.Although the accuracy of single point monitoring is not high,a large number of points extracted from the four main profiles of subsidence DEM have certain robustness.The relative error between the maximum subsidence value obtained by curve fitting and the measured maximum subsidence value is not more than20% when the maximum relative error is the largest,and the minimum relative error can reach0.7%.UAV subsidence DEM monitoring can achieve decimeter-level accuracy,and the accuracy of monitoring the maximum subsidence value is high,which provides a good data reference for mining subsidence monitoring and subsequent ecological environment restoration.