基于SBAS-InSAR技术与LSTM模型的徐州市地铁3号线地表沉降分析  

Surface subsidence analysis of metro line 3 in Xuzhou based onSBAS-InSAR technology and LSTM model

作　　者：郦嘉辉 胡晋山 岳佳乐 康建荣 黄郦 Li Jiahui;Hu Jinshan;Yue Jiale;Kang Jianrong;Huang Li(School of Geography,Geomatics&Planning,Jiangsu Normal University,Xuzhou 221116,Jiangsu,China;Nanjing Aerospace Hongtu Information Technology Co.,Ltd.,Nanjing 210000,Jiangsu,China)

机构地区：[1]江苏师范大学地理测绘与城乡规划学院,江苏徐州221116 [2]南京航天宏图信息技术有限公司,江苏南京210000

出　　处：《江苏师范大学学报（自然科学版）》2023年第4期24-29,共6页Journal of Jiangsu Normal University：Natural Science Edition

基　　金：国家自然科学基金面上项目(41671395、52074133);江苏师范大学研究生科研与实践创新计划项目(2021XKT0135);江苏师范大学研究生实践基地建设项目(Y2021GZZ0402)。

摘　　要：针对城市地铁修建过程中产生的地表沉降,基于Sentinel-1A影像数据,利用SBAS-InSAR技术反演徐州市地铁3号线从施工到运营期间沿线的地表沉降信息,并与水准观测数据进行对比.同时,将SBAS-InSAR反演得到的时序累计沉降量作为训练样本,通过长短时记忆(LSTM)神经网络对地面沉降作出预测.结果显示:SBAS-InSAR反演地表沉降量与水准观测值平均绝对误差为2.9 mm,整体误差在1 cm之内;LSTM模型对地表沉降预测的平均绝对误差大多在1.0 mm左右,表明LSTM神经网络模型可以辅助SBAS-InSAR数据预测城市地下工程建设过程中的地表沉降.In view of the surface subsidence in the process of metro construction,based on Sentinel-1A image data,SBAS-InSAR technology is applied to reverse the surface subsidence information along the metro line which is from the construction of metro line 3 in Xuzhou to normal operation,some temporary benchmarks subsidence data are compared between leveling observation and SBAS-InSAR.The accumulated subsidence amount of InSAR inversion is used as the training sample to predict the land subsidence through the LSTM.The results show that the mean absolute error of the surface subsidence value between the measured leveling and SBAS-InSAR is 2.9 mm,and the overall error is within 1 cm,and the mean absolute error of the LSTM prediction is mostly about 1.0 mm.The research results show that the LSTM neural network model can be assisted SBAS-InSAR data to predict the surface subsidence during urban underground excavation.

关 键 词：地表沉降 SBAS-InSAR LSTM神经网络 地铁工程 

分 类 号：P208[天文地球—地图制图学与地理信息工程]