盾构隧道下穿既有立交桥桥墩沉降时间序列预测研究  

Research on Bridge Pier Settlement Time Series Prediction of Shield Tunnel Through the Existing Overpass Bridge

作　　者：王杰伟 Wang Jiewei(China Railway 18th Bureau Group Municipal Engineering Co.,Ltd.,Tianjin 300222,China)

机构地区：[1]中铁十八局集团市政工程有限公司,天津300222

出　　处：《市政技术》2025年第3期224-232,共9页Journal of Municipal Technology

摘　　要：为解决城市敏感建筑物沉降变形预测中单一因素分析导致的不准确性,依托乌鲁木齐市轨道交通2号线中桥站—碾子沟站区间项目,开展了盾构下穿既有立交桥桥墩沉降变形规律时序预测研究;采用盾构掘进参数、地层物理力学参数以及沉降监测数据作为模型输入变量,通过BO优化算法对GRU模型超参数进行了全局优化,实现了盾构下穿条件下对桥墩沉降情况的准确预测。分析结果表明,BO-GRU模型的桥墩整体沉降预测趋势与实际桥墩沉降变形一致,相比于GRU模型,RMSE降低了74.69%,MAPE降低了54.16%,R2提升了1.64%,模型预测精度得到了显著提升。该研究过程克服了沉降分析中单一因素的局限性,涵盖了施工过程中沉降影响信息,减小了因地表沉降带来的风险。In order to solve the inaccuracy caused by single factor analysis in the prediction of settlement and deformation of urban sensitive buildings,the Urumqi Railway Transit Line 2 Zhongqiao Station to Mizigou Station Interval Project is taken as an example to carry out a study on the prediction of the settlement and deformation law of the piers of the existing overpasses under the shield structure;The shield tunneling parameters and the physical and mechanical parameters of the stratum is the input variables.The hyper-parameters of the GRU model are optimized by the BO optimization algorithm so as to achieve an accurate prediction of the settlement of the piers of the shield through the bridge.The analysis results show that the overall settlement prediction trend of the BO-GRU model abutment is consistent with the actual one.Compared with the GRU model,the RMSE is reduced by 74.69%,the MAPE index is reduced by 54.16%,and the R2 index is improved by 1.48%,and the model prediction performance has been significantly improved.The study overcomes the limitation of single-factor in settlement analysis,covers the information of settlement impacts during the construction process,and reduces the risk by surface settlement.

关 键 词：盾构隧道 桥墩沉降 贝叶斯优化 深度学习 

分 类 号：U443.22[建筑科学—桥梁与隧道工程] U446[交通运输工程—道路与铁道工程]