摘要
掌子面溜塌不仅会导致设备损坏和人员伤亡,还可能引发重大工程事故,严重威胁隧道安全施工与运营。准确预测掌子面的溜塌行为并评估其坍塌风险,对于隧道工程的设计、施工和管理具有重要意义。本文采用融合算法对掌子面位移进行了预测并将预测结果与其他算法进行了对比,而后采用D-S风险评价方法对掌子面坍塌风险进行了评估,结果表明:不同算法平方相关R^(2)的计算值均超过了0.9,预测值与真实值之间均拟合程度较高;RMSE的计算结果均偏大,但XGBoost算法及LSTM算法的均方根误差都超过了3,而融合算法的计算结果均小于2;XGBoost算法、LSTM算法及融合算法掌子面位移预测结果与样本值均出现了不同程度的误差,但XGBoost算法、LSTM算法的预测结果相比于融合算法的预测结果离散程度均较高;在评估的5个断面中有两个断面坍塌风险等级为Ⅰ级,两个断面坍塌风险等级为Ⅱ级,DK1+800断面坍塌风险等级为Ⅲ级,施工时要加强防控。
The collapse of the tunnel face not only causes equipment damage and personnel injury,but also may lead to major engineering accidents,seriously threatening the safe construction and operation of tunnels.Accurately predicting the collapse behavior of the tunnel face and evaluating its collapse risk is of great significance for the design,construction,and management of tunnel engineering.This article uses a fusion algorithm to predict the displacement of the tunnel face and compares the prediction results with other algorithms.Then,the D-S risk assessment method is used to evaluate the collapse risk of tunnel face.The results show that the calculated values of the squared correlation R^(2) of different algorithms all exceed 0.9,and the fitting degree between the predicted values and the true values is relatively high.The calculation results of RMSE are relatively large,but the root mean square errors of XGBoost algorithm and LSTM algorithm exceed 3,while the calculation results of fusion algorithm are less than 2.The prediction results of tunnel face displacement using XGBoost algorithm,LSTM algorithm,and fusion algorithm all show varying degrees of error compared to the sample values.However,the prediction results of XGBoost algorithm and LSTM algorithm have a higher degree of dispersion compared to the prediction results of fusion algorithm.Among the 5 evaluated sections,two sections have a collapse risk level of level I,two sections have a collapse risk level of levelⅡ,and the DK1+800 section has a collapse risk level of levelⅢ.During construction,prevention and control measures should be strengthened.
作者
刘治国
LIU Zhiguo(China Railway 11th Bureau Group No.4 Engineering Co.Ltd.,Wuhan Hubei 430070,China)
出处
《铁道建筑技术》
2024年第11期146-150,共5页
Railway Construction Technology
基金
湖北省技术创新专项任务(重大项目)(2022BEC002)。
关键词
隧道开挖
掌子面
融合算法
位移预测
风险评估
tunnel excavation
tunnel face
fusion algorithm
displacement prediction
risk assessment