基于有限元模拟与神经网络的隧道反演分析研究  

Inversion Analysis Method of Tunnel on Basis of Finite Element Analysis and Neural Network

作　　者：韩桂武[1] 常怀民[1] 杨威[1] 刘刚[1] 

机构地区：[1]中国石油天然气管道工程有限公司,河北廊坊065000

出　　处：《施工技术》2008年第9期85-87,共3页Construction Technology

基　　金：国家自然科学基金资助项目(50474018);国家自然科学基金资助项目(50490274)

摘　　要：依据某隧道施工设计及现场施工建立线弹性的隧道有限元模型,描述围岩位移场的空间分布。分析了隧道埋深、围岩性质、衬砌厚度、开挖方法及空间效应对监测点位移的影响情况。在此基础上建立围岩力学参数的反演分析流程,并建立位移与围岩力学参数的神经网络关系,最后用工程实例来证明该反演方法的可行性和有效性。According to construction, design and on-site conditions, authors firstly set up finite element model of a tunnel, and then describe the distribution of surrounding rock displacement. Secondly the causes that affect the surrounding rock displacement such as buried depth, rock mechanical parameters, liner thickness, excavation style and special effect are studied. All above factors come to an inversion analysis method with a flow process chart, and the relationships between mechanical parameters and surrounding rock displacements are simulated by neural network. Finally this method is used to deduce mechanical parameters in special area, and the result proves feasibility and availability of inversion method.

关 键 词：隧道工程 反演分析 围岩位移 神经网络 有限元分析 

分 类 号：U452[建筑科学—桥梁与隧道工程]