摘要
在深基坑开挖模拟过程中有限元数值解法比传统计算方法有较大的优势。有限元计算前要解决本构模型的选择及土层参数的取值问题,采用比较适合工程实际的Mohr-Coulomb模型。为了准确测定淤泥土层相关参数数值,采用BP神经网络进行反演分析,使神经网络的非线性映射功能得到充分体现。采用分模块处理方法作为神经网络结构,使网络能够顺利收敛。通过某工程深基坑设计参数的反演分析计算,再利用反算参数正算来预测工程实际情况的支护结构变形值。证明了该系统进行土层参数分析的合理性。
In the simulation process of deep foundation excavation,the finite element numerical method has more advantages than the traditional method.Before finite element calculation,the choice of constitutive model and the value of soil layer parameters should be solved.Mohr-Coulomb model,which is more suitable for engineering practice is adopted.In order to accurately determine soil parameters,the BP neural network is adopted to improve the inversion analysis of nonlinear mapping function of neural network are fully reflected.The neural network structure is divided into modules to make the network converge smoothly.Based on the inversion analysis and calculation of the design parameters of deep foundation excavation in a certain project,the deformation value of the support structure in the actual situation of the project is predicted by the forward calculation of the inverse calculation parameters.It is proved that this system is reasonable for soil parameter analysis.
作者
刘超
邹玉强
赵志达
李维娜
LIU Chao;ZOU Yuqiang;ZHAO Zhida;LI Weina(Wuhan Tunnel Rail Transit Engineering Technology Co.,Ltd.,Hubei,Wuhan 430080,China)
出处
《施工技术》
CAS
2020年第13期56-60,共5页
Construction Technology
基金
武汉市市政建设集团有限公司科研项目(wszky201821)。
关键词
地铁
深基坑
设计参数
BP神经网络
反演分析
变形预测
subways
deep foundation excavation
design parameters
BP neural network
back analysis
deformation prediction
