• 中国科学引文数据库(CSCD)来源期刊
  • 中文核心期刊中文科技核心期刊
  • Scopus RCCSE中国核心学术期刊
  • 美国EBSCO数据库 俄罗斯《文摘杂志》
  • 《日本科学技术振兴机构数据库(中国)》
二维码

隧道建设(中英文) ›› 2019, Vol. 39 ›› Issue (S1): 220-226.DOI: 10.3973/j.issn.2096-4498.2019.S1.031

• 研究与探索 • 上一篇    下一篇

基于BP神经网络的阿拉套山隧道围岩物理力学参数反演分析

赵俊杰1, 贾斌2, 张东2, 曹明星1, 李德武1, *   

  1. (1. 兰州交通大学土木工程学院, 甘肃兰州 730070; 2. 中铁七局集团有限公司, 河南郑州 450000)
  • 收稿日期:2019-03-26 出版日期:2019-08-30 发布日期:2019-09-12
  • 作者简介:赵俊杰(1993—),男,甘肃甘谷人,兰州交通大学桥梁与隧道工程专业在读硕士,研究方向为隧道工程爆坡振动及支护受力分析。Email: 1336219071@qq.com。*通信作者: 李德武, Email: lidewu1965@163.com。
  • 基金资助:

    中铁七局2018年科技发展计划项目

Inverse Analysis of Physical and Mechanical Parameters of Alataoshan Tunnel Surrounding Rock Based on BP Neural Network

ZHAO Junjie1, JIA Bin2, ZHANG Dong2, CAO Mingxing1, LI Dewu1, *   

  1. (1. School of Civil Engineering, Lanzhou Jiaotong University, Lanzhou 730070, Gansu, China; 2. China Railway Seventh Bureau Group Co., Ltd., Zhengzhou 450000, Henan, China)
  • Received:2019-03-26 Online:2019-08-30 Published:2019-09-12

摘要:

为研究高寒地区设置中心深埋水沟单线铁路隧道的围岩力学参数问题,以兰新铁路新建博州支线阿拉套山隧道为工程背景,基于FLAC3D数值模拟软件联合MATLAB中的神经网络工具箱构建BP神经网络算法,建立隧道开挖位移正演和反演模型,对围岩物理力学参数作反演分析。通过对中心水沟开挖前的拱顶和拱腰监测数据做拟合分析,发现隧道变形已趋于稳定,反演过程不需考虑中心水沟开挖对围岩的二次扰动。以水沟开挖前的拱顶沉降值和拱腰收敛值作为输入函数,以围岩的体积弹性模量K、剪切弹性模量G、黏聚力c、内摩擦角φ、重度γ作为输出函数训练神经网络模型,利用训练好的模型进行所需参数的反演分析。将反演参数代入FLAC3D正演模型计算后,提取中心水沟开挖前的拱顶沉降值和拱腰收敛值,与中心深埋水沟开挖前的实际监控量测值相比较为接近。结果证明,围岩物理力学参数的反演较为合理,对于变形的预测较为准确,可为隧道后期工程的施工和优化设计提供参考。

关键词: 隧道, 围岩位移, BP神经网络, 中心深埋水沟, 反演分析, 物理力学参数

Abstract:

In order to study the surrounding rock mechanical parameters of the singletrack railway tunnel in the alpine region with central deepburied ditch. Taking Alatao Mountain Tunnel of the newly built Bozhou branch line as the engineering background, based on BP neural network algorithm combined with the FLAC3D numerical simulation software and neural network toolbox in MATLAB. BP neural network algorithm is used to establish forward and inversion models of tunnel excavation displacement, and the inversion analysis of the physical and mechanical parameters of the surrounding rock is carried out. By fitting the tunnel vault and the tunnel arch monitoring data before excavation of the central deepburied ditch, it is found that the tunnel deformation has become stable, and inversion process does not need to consider the secondary disturbance of the surrounding rock in the excavation of center deepburied ditch. The vault settlement value and the arch convergence value before the excavation are taken as input function, and the bulk elastic modulus (K), shear elastic modulus (G), cohesion (c), internal friction of the surrounding rock angle (φ) and the severity (γ) are used as output functions to train the neural network model, and the trained model is used to perform the inverse analysis of the required parameters. After substituting the inversion parameters into the FLAC3D forward model, the settlement value and the arch convergence value of the vault before excavation are extracted, which is close to the actual monitoring value before the excavation of the central deepburied ditch. The results show that the inversion of physical and mechanical parameters of surrounding rock is reasonable, and the prediction of deformation is accurate, which provide reference for the construction and optimization design of subsequent works.

Key words: surrounding rock displacement, BP neural network, central deepburied ditch, inversion analysis, tunnel physical and mechanical parameters

中图分类号: