基于层次聚类算法的静力触探试验土体分类方法及试验研究
DOI:
作者:
作者单位:

作者简介:

通讯作者:

中图分类号:

P642

基金项目:

地调局大调查项目(GZH20160060202)资助.


Soil Classification Method and Experimental Research from CPT based on Hierarchical Clustering Algorithm
Author:
Affiliation:

Fund Project:

The GZH20160060202 Project is funded by the Bureau of Geological Survey.

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    划分土层、辨别土类是静力触探(CPT)成果应用的基础。常规的人工分层效果差强人意,而土体行为分类法尽管可靠性高,但无法起到分层效果。引用层次聚类算法,通过对变量的选择、数据的标准化、距离矩阵的生成和类数目的确定,得到了基于层次聚类算法的CPT土体分类流程图。采用自主研发的静力触探-钻探一体机,在汉江一级阶地和长江一级阶地上展开试验,利用层次聚类算法对地层土体进行划分,将土层划分结果与钻孔柱状图展开对比分析,结果表明:以qc-fs和qc-Rf为初始参数的聚类分层图均能够较准确识别主层的位置,其中,以qc-Rf为初始参数的聚类分层结果比qc-fs更准确,能够识别更多的次要层以及钻孔柱状图无法体现的次要层、过渡层和薄夹层,但是无法判断土层的具体类别以及单一的类(离群值)到底是属于过度带还是异常值。建议在后期研究中将孔隙水压力纳入聚类分析中,研究孔隙水压力对聚类分层效果的影响。同时,将聚类分层图与土体行为分类法结合起来,达到划分土层、辨别土类、细化土层和识别异常地层的目的。

    Abstract:

    Dividing the soil layer and identifying soil types is the basis for the application of CPT. The conventional artificial stratification method is not satisfactory, but the soil behavior classification method cannot achieve the stratification effect despite its high reliability. By using hierarchical clustering algorithm, the flow chart of CPT soil classification based on hierarchical clustering algorithm was obtained through the selection of variables, the standardization of data, the generation of distance matrix and the determination of the number of classes. Using the self-developed static cone penetration-drilling integrated machine, experiments were carried out on the first terrace of the Han-jiang River and the first terrace of the Yangtze River. The stratigraphic soils were divided by hierarchical clustering algorithm, and the results of soil stratification were compared with the borehole histogram. The results show that the clustering layers with qc-fs and qc-Rf as the initial parameters can identify the location of the main layer more accurately, and the results with qc-Rf as the initial parameters are more accurate than those with qc-fs, which can identify more secondary layers and the secondary layers, transition layers and thin layers that cannot be represented by the borehole histogram. However, it is impossible to determine the specific types of soil and whether those single classes (outliers) belong to excessive bands or abnormal values. It is suggested that the pore water pressure should be included in the cluster analysis to study the effect of pore water pressure on the clustering stratification. At the same time, the clustering and stratification map is combined with the soil behavior classification method to classify the soil layer, distinguish the soil type, refine the soil layer and identify the abnormal stratum.

    参考文献
    相似文献
    引证文献
引用本文

吕树胜,陈培帅,邱敏,等. 基于层次聚类算法的静力触探试验土体分类方法及试验研究[J]. 科学技术与工程, 2021, 21(7): 2609-2615.
Lü Shusheng, Chen Peishuai, Qiu Min, et al. Soil Classification Method and Experimental Research from CPT based on Hierarchical Clustering Algorithm[J]. Science Technology and Engineering,2021,21(7):2609-2615.

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2019-06-03
  • 最后修改日期:2021-02-23
  • 录用日期:2019-09-03
  • 在线发布日期: 2021-03-31
  • 出版日期:
×
律回春渐,新元肇启|《科学技术与工程》编辑部恭祝新岁!
亟待确认版面费归属稿件,敬请作者关注