隧道软弱围岩的改进BP神经网络位移反分析  

Displacement Back Analysis of Tunnels in Soft and Weak Rocks Based on Improved BP Neural Network Method

作　　者：高攀科[1,2] 谢永利[1] 

机构地区：[1]长安大学公路学院,陕西西安710064 [2]陕西铁路工程职业技术学院,陕西渭南714099

出　　处：《郑州大学学报（工学版）》2013年第1期23-26,共4页Journal of Zhengzhou University（Engineering Science）

基　　金：国家交通部西部交通建设科技项目(20073180038);陕西省教育厅资助项目(12JK0926)

摘　　要：基于一般BP神经网络特点,特别是在缺陷认识的基础上,采用改进BP神经网络对隧道软弱围岩进行位移反分析.通过增加神经元输出反馈量、运用二分法原理确定隐层单元数和改进惯性校正法、变步长算法、改进误差函数等方法,分别从BP神经网络结构和算法两个方面进行改进.并将现场监测数值和反分析计算位移值进行比较,其综合相对误差均控制在4%以内,取得了良好效果.成功将有限元数值仿真与BP神经网络原理结合,可为软弱围岩隧道设计和施工提供准确可靠的参数信息.Based on the realization of the defects in the general BP（ error back propagation algorithm）artificial neural network（ANN）,an approach based on improved BP ANN for displacement back analysis of soft and weak rocks for tunnels is proposed in this paper. This BP ANN was improved in two ways,neural structure and algorithm, by using increased feedback in output of neural cell, dichotomy theory to chose neurals＇ number of hidden layer, improved inertia revise, algorithm, varying step-size algorithm, improved error function. The in- tegrated error between monitor number and simulation number was controled with in 4%. Aiming at the exact and dependable parameters for construction of tunnels with soft and weak rocks, BP ANN was combined with FEM （finite element modeling） numerical simulation successfully.

关 键 词：隧道工程 神经网络 反分析 位移 软弱围岩 

分 类 号：U451[建筑科学—桥梁与隧道工程]