基于稀疏字典学习的路堤长期沉降预测研究

周泽灵; 沈坚; 戴显荣; 段冰; 潘晓东

doi:10.3724/j.gyjzG24090604

基于稀疏字典学习的路堤长期沉降预测研究

doi: 10.3724/j.gyjzG24090604

周泽灵^1,2,
沈坚³,
戴显荣³,
段冰³,
潘晓东^1,2

1. 浙江工业大学土木工程学院, 杭州 310014;
2. 浙江工业大学全省滨海基础设施绿色建造与智能运维重点;
实验室, 杭州 310014;
3. 浙江数智交院科技股份有限公司, 杭州 310030

基金项目:

浙江省交通运输重大研发项目（ZJXL-JTT-202201A）。

详细信息

作者简介:
周泽灵，硕士研究生，主要从事岩土工程的研究。电子信箱:zhouzeling@zjut.edu.cn

计量
- 文章访问数: 28
- HTML全文浏览量: 10
- PDF下载量: 0
- 被引次数: 0
出版历程
- 收稿日期: 2024-09-06
- 网络出版日期: 2025-11-05

Embankment Long-Term Settlement Prediction Based on Sparse Dictionary Learning

1. College of Civil Engineering, Zhejiang University of Technology, Hangzhou 310014, China;
2. Zhejiang Key Laboratory of Green Construction and Intelligent Operation & Maintenance for Coastal Infrastructure(Zhejiang University of;
Technology), Hangzhou 310014, China;
3. Zhejiang Institute of Communications Co., Ltd., Hangzhou 310030, China

摘要

摘要: 准确预测路堤长期沉降有利于保障道路安全和正常运营。现有研究多根据监测数据对土体参数进行反分析，利用反分析所得土体参数来更新路堤后续沉降计算值，此方法因计算成本高而难以在实际工程中广泛运用。基于此，开展了基于稀疏字典学习的路堤长期沉降预测研究，采用有限元数值计算结果构建字典，通过路堤沉降和水平位移监测数据识别字典中的重要原子并计算其权重，基于少量重要原子的加权线性组合来预测长期沉降。以澳大利亚Ballina试验路堤为例来说明和验证方法的有效性。结果表明，该方法可从有限元数值计算所得字典中识别出重要原子及其权重，通过稀疏字典学习结合多源监测数据可准确预测路堤的长期沉降，计算成本低且预测准确性高。
- 稀疏字典学习 /
- 多源监测数据 /
- 正交匹配追踪 /
- 沉降预测 /
- 有限元模型
Abstract: An accurate prediction of long-term embankment settlements is crucial for ensuring road safety and maintaining normal operations. Existing studies have employed inverse analysis of soil parameters based on monitoring data， subsequently updating settlement predictions using the inferred parameters. However， this method is hindered by high computational costs， limiting its widespread application in practical engineering. This paper proposed a long-term settlement prediction method for embankments based on sparse dictionary learning. A dictionary was constructed using finite element simulation results， and key atoms within the dictionary were identified and weighted through the analysis of settlement and horizontal displacement monitoring data. The long-term settlement was then predicted as a linear combination of a few significant atoms. The effectiveness of the proposed method was demonstrated using the Ballina trial embankment in Australia. Results indicated that the method successfully identified key atoms and their weights based on the dictionary derived from finite element analysis. By integrating sparse dictionary learning with multi-source monitoring data， this approach facilitated accurate long-term settlement predictions with low computational cost and high prediction accuracy.
- sparse dictionary learning /
- multi-source monitoring data /
- orthogonal matching pursuit /
- settlement prediction /
- finite element model

HTML全文

参考文献(16)

[1]	程健. 软土地基路堤工后沉降时参反演与预测[D]. 杭州:浙江大学，2005.
[2]	潘敏，邓志平，蒋水华. 基于边界模型和广义耦合马尔可夫链模型的地层变异性模拟方法[J]. 地质科技通报，2022，41（2）:176-186.
[3]	KELLY R，HUANG J. Bayesian updating for one-dimensional consolidation measurements[J]. Canadian Geotechnical Journal，2015，52（9）:1318-1330.
[4]	陶袁钦. 基于贝叶斯理论的岩土参数概率反分析与变形预测方法[D]. 杭州:浙江大学，2022.
[5]	陶袁钦，孙宏磊，蔡袁强. 考虑约束的贝叶斯概率反演方法[J]. 岩土工程学报，2021，43（10）:1878-1886.
[6]	TIAN H M，WANG Y. Data-driven and physics-informed Bayesian learning of spatiotemporally varying consolidation settlement from sparse site investigation and settlement monitoring data[J]. Computers and Geotechnics，2023，157，105328.
[7]	TIAN H M，WANG Y，PHOON K K. Real-time fusion of multi-source monitoring data with geotechnical numerical model results using data-driven and physics-informed sparse dictionary learning[J]. Canadian Geotechnical Journal，2024，61（11）.
[8]	谢家新. 稀疏信号恢复问题的几类算法及应用研究[D]. 长沙:湖南大学，2017.
[9]	USMAN K. Introduction to Orthogonal Matching Pursuit[G]. 2017.
[10]	TIAN H M，WANG Y. Optimal selection of dictionary atoms for sparse dictionary learning of time-varying monitoring data in two-dimensional geotechnical problems[J]. Computers and Geotechnics，2024，165（1），105953.
[11]	KELLY R，SLOAN S，PINEDA J，et al. Outcomes of the Newcastle symposium for the prediction of embankment behaviour on soft soil[J]. Computers and Geotechnics，2018，93:9-41.
[12]	JOSTAD P H，PALMIERI F，ANDRESEN L，et al. Numerical prediction and back-calculation of time-dependent behaviour of Ballina test embankment[J]. Computers and Geotechnics，2018，93:123-132.
[13]	PHOON K K，KULHAWY F H. Evaluation of geotechnical property variability[J]. Canadian Geotechnical Journal，1999，36（4）:625-639.
[14]	PHOON K K，KULHAWY F H. Characterization of geotechnical variability[J]. Canadian Geotechnical Journal，1999，36（4）:612-624.
[15]	邓志平，钟敏，潘敏，等. 考虑参数空间变异性和基于高效代理模型的边坡可靠度分析[J]. 岩土工程学报，2024，46（2）:273-281.
[16]	DOHERTY J，GOURVENEC S，GAONE F，et al. A novel web-based application for storing，managing and sharing geotechnical data，illustrated using the national soft soil field testing facility in Ballina，Australia[J]. Computers and Geotechnics，2018，93:3-8.