Machine learning-based prediction and interpretability analysis of Longxi loess collapsibility coefficient

LIU De-ren; WANG Kai-qiang; ZHANG Yan-jie; WANG Shuai-qun; WANG Yu-fei

doi:10.16285/j.rsm.2025.0588

English Version Journal Login

您当前的位置：

首页 >

文章列表页 >

Machine learning-based prediction and interpretability analysis of Longxi loess collapsibility coefficient

更新时间：2026-08-04

- Machine learning-based prediction and interpretability analysis of Longxi loess collapsibility coefficient
- Vol. 47, Issue 7, (2026)
- 作者机构：
  
  兰州交通大学土木工程学院,甘肃,兰州,730070
- 作者简介：
- 基金信息：
- DOI：10.16285/j.rsm.2025.0588
  CLC：
- Online First：08 July 2026，
  
  Published：2026
- 稿件说明：
移动端阅览
LIU De-ren, WANG Kai-qiang, ZHANG Yan-jie, et al. Machine learning-based prediction and interpretability analysis of Longxi loess collapsibility coefficient[J]. 2026, 47(7).
DOI：

LIU De-ren, WANG Kai-qiang, ZHANG Yan-jie, et al. Machine learning-based prediction and interpretability analysis of Longxi loess collapsibility coefficient[J]. 2026, 47(7). DOI： 10.16285/j.rsm.2025.0588.

摘要

黄土湿陷系数预测方法以回归分析为主，但受黄土区域性影响显著，预测准确度不足。现有的机器学习预测方法存在解释性局限。针对陇西地区黄土，通过相关性与聚类分析筛选密度、饱和度、自重应力、液限和内摩擦角5个关键影响因素，基于极端梯度提升树（extreme gradient boosting，简称XGBoost）、随机森林（random forest，简称RF）、支持向量机（support vector machine，简称SVM）和回归分析方法构建了黄土湿陷系数预测模型，并结合沙普利加性解释（Shapley additive explanations，简称SHAP）技术对预测结果进行了可解释性分析。研究结果表明：陇西黄土湿陷系数与密度、干密度、孔隙比、饱和度、内摩擦角等指标的相关系数在0.63～0.90之间，具有极强或强的相关性；XGBoost模型和RF模型的湿陷系数及湿陷性等级预测准确率均超90%，性能优于SVM模型和回归方法。SHAP分析揭示，密度小于1.44 g/cm

的黄土湿陷概率较高，饱和度大于29.0%的黄土湿陷等级相对较低，且处于高饱和度、高自重应力状态下的黄土湿陷倾向性更弱，湿陷性等级更低。研究结果可为陇西地区黄土工程建设提供理论参考。

Abstract

关键词

Keywords

references

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

A review of the application of artificial intelligence in underground engineering for compressed air energy storage

Intelligent geological condition recognition in shield tunneling via time-series clustering and online learning

Slope angle intelligent design based on Gaussian process with combinatorial kernel function

Application of Gaussian process machine learning to slope stability evaluation

Discussion on relationship between structural strength and collapsibility of loess

Related Author

GE Xin-bo

HUANG Jun

ZHAO Tong-bin

TAO Gang

MA Hong-ling

WANG Wei

DUAN Yue-qiang

LI Shuang

Related Institution

School of Civil Engineering and Architecture, Zhejiang Sci-Tech University

State Key Laboratory of Geomechanics and Geotechnical Engineering Safety, Institute of Rock and Soil Mechanics, Chinese Academy of Sciences

China Energy New Energy Storage Technology (Shandong) Co., Ltd.

College of Energy and Mining Engineering, Shandong University of Science and Technology

Xiamen Engineering Co. Ltd. of CCCC First Highway Engineering Co. Ltd.

AI问答

⁰