基于VMD-GRU的地铁隧道台阶法施工地表沉降预测  

Surface settlement prediction of subway tunnels constructed by step method based on VMD-GRU

作　　者：李涛[1] 杨腾宇 刘波[1,2] 陈前 LI Tao;YANG Tengyu;LIU Bo;CHEN Qian(School of Mechanics and Civil Engineering,China University of Mining and Technology(Beijing),Beijing 100083,China;State Key Laboratory for Geomechanics and Deep Underground Engineering,Beijing 100083,China)

机构地区：[1]中国矿业大学(北京)力学与建筑工程学院,北京100083 [2]深部岩土力学与地下工程国家重点实验室,北京100083

出　　处：《华中科技大学学报（自然科学版）》2023年第7期48-54,62,共8页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金青年基金资助项目(51508556);越崎青年学者资助项目(800015z1166);中央高校基本科研业务费专项资金资助项目(2020YJSLJ14).

摘　　要：为更精确预测施工过程中的地表沉降值变化,提出了基于变分模态分解(VMD)与门控循环单元神经网络(GRU)的地表沉降预测方法.使用VMD将地表沉降数据分解为本征模态函数(IMF),再用GRU对IMF进行预测,叠加预测分量,输出预测结果.以北京市11号线西段金顶街至金安桥区间台阶法施工地铁隧道的地表沉降实测数据为算例进行分析,结果表明:在与实测数据差值1 mm置信区间上置信度达到98.7%,对短期突变数据预测较好,在中期沉降阶段预测精度较高,能够进行沉降风险预警,具有较高的预测准确率与实用性.In order to more accurately predict the changes of surface settlement values during the construction process,a surface settlement prediction method based on variational modal decomposition(VMD)and gated recurrent unit neural network(GRU)was proposed.The method useed VMD to decompose the surface settlement data into an intrinsic modal function(IMF),and then GRU to predicted the IMF,superimposed the predicted components,and outputed the prediction results.Finally,the measured surface settlement data of the subway tunnel constructed by the step method in the west section of Beijing Line 11 between Jindingjie station and Jinanqiao station was used as an example for analysis.The results showed that the confidence level reaches 98.7%within a confidence interval of 1 mm difference from the measured data,which is good for predicting short-term sudden changes.In the midterm settlement stage,the prediction accuracy is high,and it can be used for settlement risk warning,with high prediction accuracy and practicality.

关 键 词：地下工程 台阶掘进法 沉降预测 变分模态分解 闸门重复单元 

分 类 号：U121[交通运输工程]