Research On Intelligent Monitoring And Evaluation Of Mine Geological Disasters

Geological disasters caused by mining not only cause serious damage to the mining environment,but also threaten the safety of people's lives and property.Over the years,scholars have carried out a lot of research on the basic theories and evaluation and prediction methods of mine geological disasters,and gradually established a relatively mature system of mine geological disaster investigation and evaluation technology and methods.However,the existing methods have problems such as single monitoring methods,poor timeliness,low accuracy and intelligence of evaluation and prediction.The thesis takes the Ningdong Coal Base,a large national energy and chemical base,as the research object.Aiming at the problems above,employing multi-source remote sensing monitoring methods such as remote sensing,unmanned aerial vehicles,airborne LIDAR and so on comprehensively,combined with artificial intelligence,big data and numerical simulation technologies to carry out research on technical methods of mine geological hazard survey,monitoring,evaluation and prediction based on multi-source remote sensing data at multiple time and space scales and integrated above and below ground,providing new technical means for sustainable development of mining areas and intelligent disaster prevention and mitigation.The main research work carried out in the thesis and the results obtained are as follows:(1)The thesis elaborated the definition and classification of geological disasters in mines systematically.The influencing factors and formation mechanism of disasters are analyzed aimed at the ground movement and deformation problems caused by the subsidence of coal mining areas.The basic theoretical framework and technical method system of.mine geological disasters in centralized coal mining areas are initially established based on multi-source remote sensing data.(2)The dynamic monitoring research of single ore body mining subsidence based on UAV-borne LIDAR technology was carried out.The numerical simulation method of ANSYS combined with FLACD was used to establish a three-dimensional numerical model of coal mining underground,and a comparative analysis based on LIDAR monitoring results and underground numerical simulation predicted settlement results was carried out to predict surface movement and deformation characteristics and changing laws under different mining progress.Finally,a set of technical method processes based on the integration of surface and underground three-dimensional mining subsidence monitoring and simulation prediction was summarized.(3)Taking the ground fissures caused by mining in key mining areas as the research object,the research on intelligent recognition method of complex geological disaster elements in mining area based on UAV aerial photos was carried out.Based on deep learning and semantic segmentation technology,U-Net,Res-U-Net,BASNet and PSPNet models are used to detect ground fissure targets.The MIOU index was used to evaluate the accuracy of the detection results,and the recognition effect of different models under different conditions was summarized,so as to provide reference for future research on intelligent recognition methods of ground fissures.(4)Based on the PCI GXL remote sensing data intelligent processing system,the automatic and fast processing method of satellite data was studied.Using human-computer interactive interpretation combined with automatic extraction technology,the long sequence dynamic monitoring method of mine geological disasters before and after mine development and construction in the study area was studied.Combined with the evaluation factors of mine geological disasters extracted by remote sensing,the index system of mine geological disaster risk evaluation is established,and the importance of evaluation factors was intelligently ranked by using XGBoost algorithm in integrated learning.XGBoost combined with AHP was used to complete the risk assessment of regional geological disasters in mines,which effectively solves the limitation of the single traditional assessment method interfered by human factors.