复杂地质条件深埋特长隧道地应力场智能反演研究  

An Intelligent Inversion Study of In-situ Stress Field in Deep Buried Super-long Tunnels under Complex Geological Conditions

作　　者：任洋 李天斌[1,2] 张佳鑫[3] 望刚炜 REN Yang;LI Tianbin;ZHANG Jiaxin;WANG Gangwei(College of Environment and Civil Engineering,Chengdu University of Technology,Chengdu 610059;State Key Laboratory of Geological Disaster Prevention and Geological Environment Protection,Chengdu University of Technology,Chengdu 610059;Guizhou Transportation Planning Survey&Design Academe Co.,Ltd.,Guiyang 550081)

机构地区：[1]成都理工大学环境与土木工程学院,成都610059 [2]成都理工大学地质灾害防治与地质环境保护国家重点实验室,成都610059 [3]贵州省交通规划勘察设计研究院股份有限公司,贵阳550081

出　　处：《现代隧道技术》2023年第2期47-53,72,共8页Modern Tunnelling Technology

基　　金：国家自然科学基金(U19A20111,42130719);四川省自然科学基金(2022NSFSC0411);地质灾害防治与地质环境保护国家重点实验室自主研究课题(SKLGP2021Z011)。

摘　　要：为提高复杂地质条件深埋特长隧道地应力反演的精度和效率,构建了一种基于构造分析+RBF(径向基函数)神经网络的复杂地质条件深埋特长隧道地应力智能反演方法,该方法的主要流程包括:首先通过区域地质条件及构造分析,建立隧址区三维地质模型,通过海姆假说和金尼克假说共同确定构造应力边界条件的范围,作为智能反演的训练样本,利用RBF神经网络方法计算寻优,获得最优应力边界条件,再以此最优应力边界条件进行地应力反演计算。选择西南地区某深埋特长隧道作为研究对象,将建立的地应力场智能反演方法应用于实际工程,并将反演计算获得的地应力值与实测地应力值进行拟合对比,其中最大主应力反演值拟合误差总体在10%左右,拟合精度近90%,说明该方法可行且反演计算结果较合理。To improve the inversion accuracy and efficiency of in-situ stress in deep-buried super-long tunnels un⁃der complex geological conditions,an intelligent inversion method is developed based on tectonic analysis and RBF(radial basis function)neural network for deep-buried super-long tunnels under complex geological conditions.The main steps of this method are as follows:firstly,the regional geological conditions and tectonic analysis will be used to establish a 3D geological model of the tunnel engineering area,and the Heim hypothesis and Kinnick hypothesis will be applied to jointly determine the range of conditions for the tectonic stress boundary as the training samples for intelligent inversion.Then,the RBF neural network method will be used to obtain the optimal stress boundary conditions,which will be used for an inversion calculation of the in-situ stress.A deep-buried super-long tunnel in southwest China is chosen as the pilot project,in which the established intelligent inversion method for in-situ stress fields is applied in an actual scenario.In addition,the in-situ stress values obtained from the inversion calcu⁃lation are fitted and compared with the actually measured values,which shows that the overall fitting error of the maximum principal stress inversion is about 10%and the fitting accuracy is close to 90%,indicating that the meth⁃od is feasible and the inversion calculation results are reasonable.

关 键 词：深埋隧道 RBF神经网络 应力边界 智能反演 地应力场 

分 类 号：U451.5[建筑科学—桥梁与隧道工程]