摘要
为提升城市复杂地层条件下深基坑降水设计的科学性与智能化水平,提出一种数值模拟与机器学习结合的基坑降水函数构建方法。以上海为研究背景,基于上海地区典型基坑降水双控分区,以⑦_(Ⅱ1-1)与⑦_(Ⅱ2-3)两个分区为例,建立三维地下水数值模型,分析了基坑面积、长宽比、开挖深度与帷幕插入目标含水层深度等因素对坑外0.5H、1H、2H、3H(H为基坑开挖深度)距离处水位降深的影响。在此基础上,采用多元线性回归方法构建显式函数关系,并进一步利用BP与MLP神经网络模型提升降深预测的精度与适用性。研究结果表明:神经网络模型对复杂变量间非线性关系具有良好拟合能力,预测值与模拟数据高度一致,可实现基坑外各监测点处水位降深的快速智能预测。本次研究为城市复杂区深基坑降水方案优化提供了新方法与技术支撑。
In order to enhance the scientific and intelligent level of deep foundation pit dewatering design,a method for constructing a foundation pit dewatering function by combining numerical simulation and machine learning was proposed taking Shanghai as the research background.Based on the typical groundwater-land subsidence dual control zoning of foundation pit dewatering in Shanghai area,a three-dimensional groundwater numerical model was established for partitions⑦_(Ⅱ1-1) and ⑦_(Ⅱ2-3).The influence of factors,including foundation pit area,aspect ratio,excavation depth,and curtain penetration depth into the aquifer on the groundwater drawdown at distances of 0.5H,1H,2H,and 3H(H is the excavation depth of the foundation pit)outside the pit were analysed.On this basis,multiple linear regression method was used to construct explicit functional relationships,and BP and MLP neural network models were utilized to improve the accuracy and applicability of drawdown prediction.The research results indicate that neural network models have good fitting ability for nonlinear relationships between complex variables,and the predicted values are highly consistent with the measured data,which can achieve rapid and intelligent prediction of groundwater drawdown at various monitoring points outside the foundation pit.The research provides new methods and technical support for optimizing deep foundation pit dewatering schemes in complex urban areas.
作者
王建秀
龙燕霞
王禹
杨天亮
殷立峰
晏殊
林子越
WANG Jianxiu;LONG Yanxia;WANG Yu;YANG Tianliang;YIN Lifeng;YAN Shu;LIN Ziyue(Tongji University,Shanghai 200092,China;Key Laboratory of Geotechnical and Underground Engineering,Ministry of Education,Shanghai 200092,China;Shanghai Institute of Natural Resources Survey and Utilization Research,Shanghai 200092,China;Shanghai Changkai Geotechnical Engineering Co.,Ltd.,Shanghai 200093,China)
出处
《上海国土资源》
2025年第3期68-78,共11页
Shanghai Land & Resources
基金
上海市住房和城乡管理委员会重点课题(2024-Z02-007)
上海长凯岩土工程有限公司课题(kh0023020231733)
上海市交通委员会科研项目(JT2025-KY-027)
上海申通地铁集团有限公司科研项目(JS-KY24R006)。
关键词
深基坑
降水预测
数值模拟
BP神经网络
MLP模型
deep foundation pit
dewatering prediction
numerical simulation
BP neural network
MLP neural network
