湖南大学学报（自然科学版）2025，Vol.52Issue(5)：57-65,9.DOI:10.16339/j.cnki.hdxbzkb.2025047
基于改进Stacking多模型融合的高速公路隧道建设碳排放预测模型
A Carbon Emission Prediction Model for Highway Tunnel Construction Based on Improved Stacking Multiple Model Integration
摘要
Abstract
Tunnel engineering,as a significant component of transportation infrastructure,has increasingly received attention for its carbon emissions during the construction process.It is a significant scientific basis for controlling carbon emissions and achieving carbon reduction in tunnel design by establishing a rational model for predicting carbon emissions in tunnel construction.Accordingly,the dataset concludes a total of 120 samples of tunnel construction carbon emissions(per meter)with different lining designs based on the construction of the Menglü highway tunnel project,considering 12 characteristic parameters,including the surrounding rock level,the total length of the tunnel,and so on.Based on the traditional Stacking algorithm,this study proposes an improved Stacking algorithm of multi-model fusion to predict carbon emissions in tunnel construction.The improved Stacking algorithm combines the various base learner training models with the residual weighting approach,which is obtained through cross-validation.Besides,the improved Stacking algorithm utilizes the original training set and the prediction results of the combined base learner as the meta-learner inputs.Therefore,the improved Stacking algorithm is not only less sensitive to noise,but also retains the original dataset information.The results demonstrate that the improved Stacking algorithm is superior to three single base learners as well as the traditional Stacking algorithm in terms of root mean square error(ERMSE),mean absolute error(EMAE),and determination coefficient(R2).Consequently,it recommends the improved Stacking algorithm for predicting carbon emissions in tunnel construction.关键词
隧道建设/碳排放/单一基学习器/Stacking模型融合/预测模型Key words
tunnel construction/carbon emissions/single base learner/stacking model integration/prediction model分类
交通运输引用本文复制引用
吴佳润,林宇亮,邢浩,宁曦..基于改进Stacking多模型融合的高速公路隧道建设碳排放预测模型[J].湖南大学学报（自然科学版）,2025,52(5):57-65,9.基金项目
国家自然科学基金资助项目(51878667),National Natural Science Foundation of China(51878667) （51878667）
湖南省自然科学基金资助项目(2021JJ30830),Natural Science Foundation of Hunan Province(2021JJ30830) （2021JJ30830）
