铁道标准设计2025，Vol.69Issue(4)：157-165,9.DOI:10.13238/j.issn.1004-2954.202212140005
基于LightGBM的盾构机姿态预测与控制研究
Research on Prediction and Control of Shield Machine Attitude Based on LightGBM
摘要
Abstract
Effectively controlling the attitude of the shield around its axis to prevent forward tilting deformation,serpentine movement,axis deviation,and correction-induced issues such as segment misalignment,cracks,and water seepage is a critical and challenging quality and safety concern in shield tunneling construction.It requires precise prediction and effective control of the shield attitude.To address this,an intelligent prediction and control method for shield attitude based on the Light Gradient Boosting Machine(LightGBM)model was proposed.Six parameters—pitch angle,roll angle,horizontal and vertical displacement of the shield cutterhead,horizontal and vertical displacement of the shield tail—were selected to describe the shield machine attitude for prediction and control.The hyperparameters of the machine learning algorithm were optimized,and an optimal LightGBM prediction model was established.Feature importance ranking was used to identify key indicators,including jack thrust,earth pressure,cutterhead torque,and tunneling speed,for optimization adjustments.Based on the LightGBM model's predictions,when significant differences between the predicted shield attitude and the design target alignment(DTA)were detected,the shield attitude could be optimized by adjusting the shield machine's operational parameters in advance.Taking the Guiyang Metro Line 3 as an example,the effectiveness of this method was validated.The research conclusions are as follows:(1)the LightGBM model can accurately predict the shield attitude control targets,with a goodness of fit(R2)exceeding 0.85.(2)The importance ranking of influencing factors for shield attitude can identify the key construction parameters to be controlled.(3)Through the optimization and adjustment of key shield construction parameters,the attitude control targets are maintained within the warning range,effectively realizing attitude control during shield tunneling construction.关键词
盾构/施工参数/预测与控制/机器学习/LightGBM算法/影响因素Key words
shield/construction parameters/prediction and control/machine learning/LightGBM algorithm/influencing factors分类
交通工程引用本文复制引用
曾铁梅,李昕懿,冯宗宝,陈虹宇,王雷,覃亚伟,徐文胜..基于LightGBM的盾构机姿态预测与控制研究[J].铁道标准设计,2025,69(4):157-165,9.基金项目
国家自然科学基金项目(51778262,71571078,51308240) （51778262,71571078,51308240）
