基于CNN-LSTM-Adaboost模型的TBM掘进参数和隧洞围岩等级预测  

Prediction of TBM excavation parameters and tunnel surrounding rock grade based on CNN-LSTM-Adaboost model

作　　者：戴明健 焦玉勇 裴成元 贾运甫[3] 梁峰[3] 张鹏[1] DAI Mingjian;JIAO Yuyong;PEI Chengyuan;JIA Yunfu;LIANG Feng;ZHANG Peng(Faculty of Engineering,China University of Geosciences(Wuhan),Wuhan 430074,China;Xinjiang Water Conservancy Development and Construction Group Co.,Ltd.,Urumqi 830000,China;Xinjiang Survey and Design Institute for Water Resources and Hydropower,Urumqi 830000,China)

机构地区：[1]中国地质大学(武汉)工程学院,湖北武汉430074 [2]新疆水发建设集团有限公司,新疆乌鲁木齐830000 [3]新疆水利水电勘测设计研究院有限公司,新疆乌鲁木齐830000

出　　处：《安全与环境工程》2025年第2期160-170,共11页Safety and Environmental Engineering

基　　金：国家自然科学基金重点项目(41920104007)。

摘　　要：硬岩隧道掘进机(TBM)在现今隧洞建设中的应用日益广泛,但是TBM深埋于地下,对地下围岩状况和掘进参数的感知能力不高,精准预测掘进参数和围岩等级对于保证TBM掘进安全具有重要意义。基于新疆某引水工程的TBM现场掘进参数和隧洞围岩地质数据,选择TBM掘进稳定段的推力、扭矩、转速、净掘进速度、施工速度、开挖比能作为模型输入参数,建立了卷积神经网络优化的长短时时序-自适应提升(CNN-LSTM-Adaboost)模型,预测各等级围岩条件下的掘进参数,并依据掘进参数数据集训练模型预测了隧洞围岩等级。结果表明:CNN-LSTM-Adaboost模型具有较高的预测精度,大部分数据的预测相对误差率(Er)在10%以内,均方根误差(RMSE)在5以内,平均绝对百分比误差(MAPE)在10%以内,拟合优度(R^(2))在0.9以上;同时,CNN-LSTM-Adaboost模型对基于掘进参数对隧洞围岩等级的识别准确率较高,综合准确率(ACC)达90%,可以应用于指导工程实践。Hard rock tunnel boring machine(TBM)is increasingly widely used in tunnel construction today,but TBM is deeply buried underground and has a low perception of the underground surrounding rock condition and excavation parameters.Accurately predicting excavation parameters and surrouding rock grade is of great significance for ensuring TBM excavation safety.Based on the on-site excavation parameters and geological data of a certain water diversion project in Xinjiang,the thrust,torque,rotational speed,net excavation speed,construction speed,and excavation specific energy of the stable section of TBM excavation were selected as input parameters for the model.A convolutional neural network optimized long short-term time sequence adaptive enhancement integrated prediction model(CNN-LSTM-Adaboost model)was established to predict excavation parameters under various levels of surrounding rock conditions.At the same time,a model was trained to predict the grade of tunnel surrounding rock based on the excavation parameter dataset.The research results show that the CNN-LSTM-Adaboost model has high prediction accuracy,with most data having a relative prediction error rate(Er)of less than 10%,a root mean square error(RMSE)of less than 5,a mean absolute percentage error(MAPE)value of less than 10%,and a goodness of fit(R^(2))value of over 0.9;meanwhile,the CNN-LSTM-Adaboost model has a high accuracy in identifying the classification of tunnel surrounding rock based on excavation parameters,with comprehensive accuracy(ACC)of 90%,which can be applied to guide engineering practice.

关 键 词：硬岩隧道掘进机(TBM) 掘进参数 掘进安全 CNN-LSTM-Adaboost模型 围岩等级 

分 类 号：X951[环境科学与工程—安全科学] TV554[水利工程—水利水电工程] U455.31[建筑科学—桥梁与隧道工程]