PDF全文阅读

王安东 ，张学钢 ， 宁   波

 （ 陕西铁路工程职业技术学院 ，陕西 渭南 714000)

摘   要 ：为有效掌握基坑变形特性，以基坑位移变形监测成果为基础，在开展基坑变形现状特征分析基础上，利用集成经验模态和样本熵分别实现变形数据的分解、重构，以准确将其分解为真实变形项和不确定变形项，再进一步通过 CRU 神经网络和多元线性回归分别实现不同变形项的预处理，实现基坑变形的高精度预测，并借其结果掌握基坑变形特性。结果表明：通过数据分解 、重构处理，能有效分解基坑变形数据，且较传统分 解思路的优越性明显；通过 AM- CRU- MLR 模型预测，预测精度相对更优，充分验证了预测思路的有效性，为基坑变形特性分析提供了一定的技术支持。

关键词：基坑 ；变形数据 ；分解处理 ；变形预测 ；变形特征

中图分类号：P642           

       文献标志码：A            

文章编号： 1005- 8249   (2024)  05- 0065- 06 

DOI:10. 19860/j.cnki.issn1005 - 8249.2024.05 .012

 

WANG Andong ,  ZHANG Xuegang ,  NING Bo

(Shaanxi Railway Institute ,  Weinan 714000 ,  China)

Abstract:In order to effectively grasp the deformation characteristics of foundation pits, based on the displacement and deformation monitoring results of foundation pits, and based on the analysis of the current deformation characteristics of foundation pits, the integrated empirical mode and sample entropy are used to decompose and reconstruct the deformation data, accurately decomposing it into real deformation terms and uncertain deformation terms. Furthermore, CRU neural network and multiple linear regression are used to preprocess different deformation terms, achieving high-precision prediction of foundation pit deformation, and using the results to grasp the deformation characteristics of foundation pits. The results shows that through data decomposition and reconstruction processing, the deformation data of foundation pits can be effectively decomposed, and its superiority is obvious compared to traditional decomposition methods; Meanwhile, the AM-CRU-MLR model showed relatively better prediction accuracy, fully verifying the effectiveness of the prediction approach and providing theoretical support for the analysis of deformation characteristics of foundation pits.

Keywords: foundation pit; deformation data; decomposition treatment; deformation prediction; deformation characteristics

作者简介： 王安东  (1992—) ,  男，硕士，工程师，研究方向 ：隧道工程，大跨度桥梁施工 、BIM 技术应用。

收稿日期：2023- 01 - 22