摘要
盾构掘进参数选取以及施工扰动控制一直是盾构工法中的关注重点,而基于实测数据挖掘理论的机器学习方法对此类问题有很好的适用性。为此,文章提出一种基于机器学习的盾构隧道施工扰动控制方法,首先借助机器学习建立地表沉降智能预测控制模型,并根据掘进参数有效预测可能发生的地表变形值;然后以地表沉降值最小为目标函数,对掘进参数进行寻优,优化获得施工扰动最小的掘进参数集。通过应用实例,采用最小二乘支持向量机和粒子群算法建立了以盾构掘进参数实时采样值为输入量、以地表沉降为输出变量的地表沉降预测控制器,基于建立的预测模型,以控制地表变形为目标对掘进参数设定值进行寻优,最终将盾构施工扰动控制在允许范围内。实践证明,此方法简单易行,能够快速分析现场采集的掘进参数,并进行施工反馈。
The selection of tunneling parameters and the control of construction disturbance have always been the focus of shield tunneling,and the machine learning method based on measured data mining theory has good applicability to such problems.This paper proposes a technique for control of the construction disturbance induced by shield tunneling based on machine learning.Firstly,an intelligent model for prediction and control of surface settlement was established by machine learning.Within the model,the possible surface settlements corresponding to certain construction parameters can be accurately predicted.Then,with minimizing surface settlement as an objective function,the optimum tunnel parameters can be identified.Furthermore,an example was presented,where the least squares support vector machine for predicting surface settlement was developed,and the parameters were optimized by adopting the PSO algorithm.In this predictor,the shield construction parameters were set as input variables while the surface settlement as output variable.Based on this model,a set of optimum matching construction parameters was found in terms of minimizing surface settlement.
作者
王鹏
WANG Peng(Nanjing Metro Construction Co.Ltd.,Nanjing 210036)
出处
《现代隧道技术》
EI
CSCD
北大核心
2019年第S02期368-373,共6页
Modern Tunnelling Technology
关键词
地铁
施工扰动控制
机器学习
盾构隧道
掘进参数
Subway
Control of construction disturbance
Machine learning
Shield tunnel
Tunnel parameters
