摘要
盾构施工过程中,现有地面沉降监测手段与方法难以满足高效安全施工要求,因此,针对盾构掘进过程中难以精确预测地面沉降问题,提出一种地面沉降量的预测方法。首先,将收集的相关数据进行预处理,预处理包含了缺失值处理、异常值处理和无量纲化处理等,预处理可以提高数据质量,从而提升降低模型训练复杂度、提升精度;然后,把处理好的数据分为训练集和测试集,对划分好的数据基于随机森林构建地面沉降影响因素分析模型,分别得出盾构机前方0环处、3环处、5环处、10环处、15环处、20环处以及盾构机后方-5环处、-10环处、-15环处、-20环处的盾构掘进参数对地面沉降的差异化影响因素及其重要度分数,并根据每环对应的不同影响因素集,进一步构建基于XGBoost地面沉降预测模型;最后,将与地面沉降相关的影响因素具体数据作为输入,多监测点的沉降量作为输出,最终得到距开挖面不同距离处的多点地面沉降量。该方法实现了地面沉降量准确预测,同时具备建模简单、准确率高等优点,对盾构施工工程及风险管控具有一定参考价值。
During the shield construction process,the existing ground subsidence monitoring means and methods could hardly meet the requirements of efficient and safe construction.Therefore,in view of the difficulties in accurately predicting the ground subsidence during the shield tunneling,this article proposes a ground subsidence prediction method.Relevant data is preprocessed first,including missing value processing,outlier processing,and dimensionless processing.Preprocessing can improve data quality,thereby reducing model training complexity and accuracy.The processed data is divided into training set and test set.Based on the random forest,an analysis model of the factors affecting land subsidence is constructed to obtain divided-data-based differential influence factors and important factors of shield tunneling parameters on ground settlement at 0 ring,3 ring,5 ring,10ring,15 ring,20 ring and behind the shield machine at-5 ring,-10 ring,-15 ring,and-20 ring.According to the different influencing factor sets corresponding to each ring,the ground subsidence prediction model based on XGBoost is further constructed.The specific data of influencing factors related to land subsidence is used as input,and the subsidence of multiple monitoring points is used as output.The ground subsidence at multiple points of different distances from the surface can realize accurate prediction of ground subsidence.At the same time,it has the advantages of simple modeling and high accuracy and great reference value for shield construction and risk management.
作者
夏汉庸
黄毅
尹和军
张良
XIA Hanyong;HUANG Yi;YIN Hejun;ZHANG Liang(Ningbo Rail Transit Group Co.,Ltd.,Ningbo 315012,China;China Railway Engineering Design and Consulting Group Co.,Ltd.,Beijing 100055,China)
出处
《铁道标准设计》
北大核心
2023年第3期140-147,共8页
Railway Standard Design
基金
宁波市轨道交通集团有限公司建设分公司科研课题(JF-ZFH-19-001)。
关键词
地面沉降预测
地面沉降影响因素
XGBoost
盾构施工
盾构掘进参数
ground subsidence forecast
factors affecting ground subsidence
XGBoost
shield construction
shield tunneling parameters
