摘要
在近接施工问题中,精准评估邻近建筑施工对既有建、构筑物的影响较为困难,且无法进行智能化预测。文章基于IFC拓展模型,建立4个包含隧道变形影响因素的拓展属性集,并结合机器学习算法及Python编程语言,提出一种邻近施工影响下既有隧道变形的智能预测方法。首先,通过数值模拟方法进行力学仿真试验和邻近施工工况模拟,得到两者对应的时序数据集以补充拓展属性集的信息;然后,通过数据处理手段,完善监测信息和场地降雨的时序数据集;最后,采用LSTM机器学习算法对4种时序数据集进行学习,以此建立完整的智能预测方法。结果表明:建立的力学仿真数据集、加卸载数据集、桩基施工数据集及场地降雨数据集对隧道变形产生不同程度的影响,基于LSTM的智能预测方法可以很好地预测邻近施工复杂环境下既有隧道的变形。
In adjacent construction issues,it is challenging to accurately assess the impact of nearby building construction on existing structures,and intelligent prediction is not feasible.This article,based on the IFC extension model,establishes four extension attribute sets that include factors influencing tunnel deformation.By combining machine learning algorithms and Python programming language,an intelligent prediction method for tunnel deformation under the impact of adjacent construction is proposed.Firstly,this article carries out mechanical simulation experiments and simulates the adjacent construction scenarios using numerical simulation,obtaining corresponding time-series datasets to supplement the information in the extension attribute sets;then,through data processing means,the monitoring data and time-series datasets of site rainfall are improved.Last but not least,the LSTM machine learning algorithm is applied to learn from the four time-series datasets,thus establishing a complete intelligent prediction method.The results show that the established mechanical simulation dataset,loading/unloading dataset,pile foundation construction dataset and site rainfall dataset have varying degrees of impact on tunnel deformation.The LSTM-based intelligent prediction method could effectively predict the deformation of existing tunnels under the complex environment of adjacent construction.
作者
安武斌
游黄斌
黄睿奕
冯萌萌
张元超
黄明
AN Wubin;YOU Huangbin;HUANG Ruiyi;FENG Mengmeng;ZHANG Yuanchao;HUANG Ming(No.5 Engineering Corporation Limited,China Railway 11th Bureau Group Corporation Limited,Chongqing 400037,China;Fuzhou University,Fuzhou Fujian 350108,China)
出处
《现代城市轨道交通》
2025年第3期18-25,共8页
Modern Urban Transit
基金
国家自然科学基金(41972276)。
关键词
地铁
智能建造
IFC模型
数值模拟
机器学习
metro
intelligent construction
IFC model
numerical simulation
machine learning
作者简介
第一作者:安武斌,男,工程师;通信作者:游黄斌,男,硕士研究生。
