节点文献

融沉系数的人工神经网络预测方法

Artificial Neural Network Forecasting Method for Thaw-Settlement Coefficient

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 姚晓亮齐吉琳

【Author】 YAO Xiao-liang,QI Ji-lin(State Key Laboratory o f Frozen Soil Engineering,Cold and Arid Regions Environmental and EngineeringResearch Institute,Chinese Academy Sciences,Lanzhou Gansu 730000,China)

【机构】 中国科学院寒区旱区环境与工程研究所冻土工程国家重点实验室

【摘要】 分析了前人关于融沉系数经验方法的研究结果,结果显示,与融沉系数关系最为密切的物性参数为液塑限、粉黏粒含量、干密度和含水量(含冰量).为了能够综合描述诸因素与融沉系数的经验关系,以兰州黄土和青藏黏土为试验对象,得到了两种具有不同物性参数的土在不同含水量和干密度条件下的融沉系数.采用BP神经网络算法对试验数据进行学习训练,得到了各因素与融沉系数间的经验关系数据库.为了提高训练样本的代表性,引用前人研究中的部分数据作为补充.对预留数据的预测结果表明,综合考虑多因素影响的BP神经网络经验方法具有较好的预测精度,而使用单一因素(含水量或干密度)预测融沉系数的经验方法其精度相对较差.

【Abstract】 Thaw-settlement is a tranditional issue in frozen soil mechanics and engineering,but so far yet there is not a widely accepted forecasting method for the thaw-settlement coefficient.Through analyzing the previous study achievement on empirical method,in this paper,Atterberg Limits,fine particles content,dry unit weight and water content are proposed to be the indexes that closely relates to thaw-settlement coefficient.In order to describe the relationship between all those factors and thaw-settlement coefficient comprehensively,Lanzhou loess and Qinghai-Tibet silty clay were taken as study objects and the thaw-settlement coefficient under different dry unit weight and water content were obtained.A database was obtained by training testing data by using BP neural network.Some data of previous research work were also trained to enhance the training sample representation.The forecasting results indicate that the BP neural network forecasting method has a better accuracy than the empirical method based on only one factor(dry unit weight or water content).

【基金】 中国科学院冻土工程国家重点实验室自主课题“冻土融化固结试验与理论研究”(2010)资助
  • 【文献出处】 冰川冻土 ,Journal of Glaciology and Geocryology , 编辑部邮箱 ,2011年04期
  • 【分类号】P642.14
  • 【被引频次】24
  • 【下载频次】301
节点文献中: 

本文链接的文献网络图示:

本文的引文网络