基于D-InSAR技术的矿区地表沉降规律分析与趋势预测  

Analysis and Trend Prediction of Surface Subsidence Patterns in Mining Areas Using D-InSAR Technology

作　　者：朱权洁[1] 谷雷 刘晓云[3] 尹永明[4] 朱斯陶 ZHU Quanjie;GU Lei;LIU Xiaoyun;YIN Yongming;ZHU Sitao(School of Emergency Technology and Management,North China Institute of Science&Technology,Sanhe 065201,China;School of Mine Safety,North China Institute of Science&Technology,Sanhe 065201,China;School of Resources and Environmental Engineering,Wuhan University of Science and Technology,Wuhan 430081,China;China Academy of Safety Science and Technology,Beijing 100012,China;School of Civil and Resources Engineering,University of Science and Technology Beijing,Beijing 100083,China)

机构地区：[1]华北科技学院应急技术与管理学院,河北三河065201 [2]华北科技学院矿山安全学院,河北三河065201 [3]武汉科技大学资源与环境工程学院,湖北武汉430081 [4]中国安全生产科学研究院,北京100012 [5]北京科技大学土木与资源工程学院,北京100083

出　　处：《金属矿山》2024年第11期125-131,共7页Metal Mine

基　　金：中央引导地方科技发展资金项目(基础研究项目)(编号:216Z5401G);河北省自然科学基金项目(编号:E2023508021);湖北省安全生产专项资金科技项目(编号:SJZX20220907);中央高校科研业务费专项(编号:3142021002);河北省研究生教育创新资助项目(编号:CXZZSS2022159)。

摘　　要：井工煤矿开采会导致地表沉陷,损坏地表建筑物、影响生态环境,甚至诱发滑坡、泥石流等灾害,对地表沉陷发展规律的准确分析和预测具有重要的现实意义。以东滩煤矿6306工作面为试验对象,选取覆盖该区域的19景Sentinel-1A影像数据,开展了基于D-InSAR技术的地表沉降规律分析与预测研究。基于“二轨法”获取了试验区域的时间序列数据,构建了SAR影像数据分析方法与处理流程;结合矿区实测数据和Sentinal-1A影像数据,验证分析了6306工作面对应的地表沉降规律;构建了基于LSTM算法的地表沉降预测模型,对比分析了LSTM、SVR和灰色GM(1,1)3种方法预测的准确性和有效性。结果表明:D-InSAR技术的监测精度高(最大误差为18.3 mm,平均差值为5.4 mm),区域广,可有效获取地表形变的时空演化规律;此外,相较于传统SVR、灰色GM(1,1)预测模型,所提出的LSTM模型平均误差约为2.98 mm,具有更高精度。Coal mining can cause surface subsidence,damage surface buildings,affect ecological environment,and even induce landslides,debris flows and other disasters,so it is of great practical significance to accurately analyze and predict the development law of surface subsidence.Taking 6306 working face of Dongtan Coal Mine as the test object,19 Sentinel-1A image data covering the area were selected to carry out the analysis and prediction of surface subsidence law based on D-InSAR technology.The main contents include:the time series data of the test area is obtained based on the"two-track method",and the analysis method and processing flow of SAR image data are constructed;Combined with the measured data of mining area and Sentinal-1A image data,the corresponding surface settlement law of 6306 working face is verified and analyzed.The accuracy and effectiveness of LSTM,SVR and grey GM(1,1)methods were compared and analyzed.The results show that the D-In-SAR technique has high monitoring accuracy(maximum error is 18.3 mm,average difference is 5.4 mm)and wide area,and can effectively obtain the spatio-temporal evolution of surface deformation.In addition,compared with the traditional SVR and grey GM(1,1)prediction model,the average error of the proposed LSTM model is about 2.98 mm,which has higher accuracy.

关 键 词：开采沉陷 D-INSAR LSTM 水准监测 趋势预测 

分 类 号：TD325[矿业工程—矿井建设]