基于BP网络的土层弹性模量反演及其在ABAQUS中的实现  

Inversion on modulus of elasticity of soil horizon based on BPnetwork and its implementation in ABAQUS

作　　者：袁三莲 伍军 YUAN Sanlian;WU Jun(CCCC Second Navigation Bureau Construction Technology Co.,Ltd.,Wuhan 430000,Hubei,China)

机构地区：[1]中交二航局建筑科技有限公司,湖北武汉430000

出　　处：《工程建设》2023年第9期13-18,共6页Engineering Construction

摘　　要：土层弹性模量一般难以通过现场勘探或室内试验得到,而开展位移反演分析具有相对优势。本文融合BP神经网络算法优化和有限元软件ABAQUS建模技术,借助ABAQUS软件中的python语言接口自行编程,以MATLAB工具箱为平台,对非均质土层的弹性模量开展位移反演分析,并预测桩基后续沉降。研究结果表明:土层弹性模量反演值与土层压缩模量地勘报告值之间存在潜在的数学函数关系;桩基沉降预测值与现场试验监测值高度吻合,并且桩基沉降预测值大于现场试验监测值。本文成果可为提高桩基沉降的预测效率,为后续的施工设计调整提供数据支撑。The modulus of elasticity of soil horizon is generally difficult to obtain through field exploration or indoor experiment,but displacement inversion analysis has comparative advantages.BP neural network algorithm optimization and finite element software ABAQUS modeling technology are integrated,the python language interface in ABAQUS software is used to program itself,and the MATLAB toolbox is used as the platform to carry out displacement inversion analysis of the modulus of elasticity of heterogeneous soil horizon,while the subsequent settlement of pile foundation is predicted.The results show that there is a potential mathematical function relationship between the inversion value of the modulus of elasticity of soil horizon and the value of the geological survey of the compressive modulus of the soil horizon.The prediction value of pile foundation settlement is highly consistent with the field test monitoring value,and the pile foundation settlement prediction value is greater than the field test monitoring value.The results can improve the prediction efficiency of pile foundation settlement and provide data support for subsequent construction design adjustment.

关 键 词：岩土工程 土层弹性模量 桩基沉降 位移反演分析 BP神经网络 ABAQUS 

分 类 号：TU473.1[建筑科学—结构工程]