基于熵权的隧道涌水量组合预测方法  

Combined predict method of tunnel water inflowbased on entropy weight

作　　者：李俊[1] 黄晓敏[1] 李睿[1] 王亮 李泽[1] LI Jun;HUANG Xiaomin;LI Rui;WANG Liang;LI Ze(School of Architecture and Engineering,Kunming University of Science and Technology,Kunming 650500,China)

机构地区：[1]昆明理工大学建筑工程学院,云南昆明650500

出　　处：《人民长江》2025年第3期140-145,共6页Yangtze River

基　　金：国家自然科学基金项目(12162018);云南交投集团公路建设有限公司项目(QT00MM202310001)。

摘　　要：隧道涌水量预测存在多种解析公式和经验公式预测方法及模型,各类公式计算结果与实际涌水量存在偏差。为了提高隧道涌水量预测公式的精度,采用熵值法计算权重,结合有序加权的方法对单项预测结果进行组合预测。基于云南省列白妥隧道实际监测隧道涌水量,选用常用的5种单项预测模型和2种组合预测模型预测,引入有效度等指标来评价不同模型预测结果。结果表明:基于熵权的组合预测方法在5个预测分段平均预测精度为96.46%,有效度为0.9419,对比最准确的单项预测结果平均精度提高18.17%,有效度提高24.6%。组合预测方法显著提升了预测精度,可为隧道排水方案的制定提供参考,降低隧道涌水带来的灾害风险。Various analytical and empirical formulas and models have been developed for predicting tunnel water inflow;however,discrepancies often exist between calculated results and actual measurements.To enhance accuracy of tunnel water inflow prediction,this study employs the entropy method to calculate weights and integrates these with the ordered weighted averaging method for combined prediction of individual model outputs.Based on the actual monitoring data of Liebaituo Tunnel in Yunnan Province,five commonly used single prediction models and two combined prediction models were evaluated based on effectiveness and other performance indicators.The results demonstrated that the entropy-based combined prediction method achieved an average accuracy of 96.46%across five prediction segments,with an effectiveness value of 0.9419.Compared with the most accurate single prediction model,this combined approach improved average accuracy by 18.17%and effectiveness by 24.6%.The combined prediction method can significantly improve prediction accuracy,provide support for formulating tunnel drainage scheme,and reduce the disaster risk caused by tunnel water inflow.

关 键 词：隧道涌水量 预测精度 解析公式法 熵值法 组合预测 有效度指标 云南省列白妥隧道 

分 类 号：U456.32[建筑科学—桥梁与隧道工程]