BP神经网络在隧道围岩力学参数反演中的应用  

Application of BP Neural Network to the Back Analysis of Mechanical Parameters of Tunnel Surrounding Rock

作　　者：文辉辉[1] 尹健民[2] 秦志光[1] 谢仁红[1] 

机构地区：[1]中交四航工程研究院有限公司交通基础工程环保与安全重点实验室,广州510230 [2]长江科学院水利部岩土力学与工程重点实验室,武汉430010

出　　处：《长江科学院院报》2013年第2期47-51,56,共6页Journal of Changjiang River Scientific Research Institute

摘　　要：以谷城至竹溪高速公路珠藏洞隧道施工监测为工程依托,根据现场变形监测数据的指数函数回归方程,对最终变形量进行了预测,并基于其预测值,借助BP神经网络的超强非线性映射能力,对隧道围岩力学参数(变形模量E、黏聚力C、内摩擦角φ)进行反演,以及时掌握开挖围岩类型和材料特性参数,为隧道工程施工和设计提供参数依据,从而达到安全施工和优化设计的目的,以实现隧道的信息化施工与设计。The aim of this research is to ensure the construction safety and optimize the design of tunnels using information technology.With the construction of Zhuzang tunnel of Gucheng-Zhuxi highway as an engineering background,we predicted the final deformation by regression equation of exponential function deduced from the field displacement measurement data.Subsequently,on the basis of the predicted deformation,we carried out back analysis on the mechanical parameters（deformation modulus E,cohesion C,internal friction angle φ） of the tunnel＇s surrounding rock through BP neural network which has good nonlinear mapping ability.The surrounding rock type and material parameters can be obtained in time to provide parameters for the design and construction of the tunnel.

关 键 词：最终变形量 BP神经网络 隧道围岩 力学参数 反演 

分 类 号：U451.2[建筑科学—桥梁与隧道工程]