中外公路2025，Vol.45Issue(1)：60-66,7.DOI:10.14048/j.issn.1671-2579.2025.01.007
基于不确定性量化理论的寒区道路服役性能预测及评价研究
Prediction and Evaluation of Road Service Performance in Cold Regions Based on Uncertainty Quantification Theory
摘要
Abstract
In cold regions,road service performance is influenced by various nonlinear dynamic factors under complex hydrothermal environments and freeze-thaw cycles.Traditional empirical and mechanical models exhibit significant limitations in comprehensively considering the impacts of complex environments and conducting quantitative analyses.This study proposed a method to predict and assess the service performance of constructed roads in cold regions.The method integrated deep neural networks with uncertainty quantification theory,addressing the inadequacies of existing models in adapting to complex environments and analyzing multiple influencing factors.By utilizing data from Montana road sections in the LTPP database,the study employed deep neural networks to predict the international roughness index(IIRI).The Morris one-at-a-time method was used to screen key influencing factors,and sensitivity analysis was conducted using the Sobol index to identify the primary factors affecting road service performance.The results indicate that road age,traffic volume,and freezing index are critical factors influencing road service performance,followed by temperature and wind speed,while annual precipitation,solar radiation,and humidity have relatively minor impacts.These findings highlight the significance of screening key factors and conducting sensitivity analyses to discern the primary and secondary factors influencing road service performance in the complex environments of cold regions.The proposed research framework effectively addresses the limitations of traditional evaluation models in adapting to complex environments and analyzing influencing factors comprehensively.It provides theoretical support and practical guidance for accurate prediction and scientific evaluation of road service performance in cold regions.关键词
寒区/道路服役性能/深度神经网络/不确定性量化/气候变化Key words
cold region/road service performance/deep neural network/uncertainty quantification/climate change分类
交通工程引用本文复制引用
刘媛媛,司君岭,郑好,任道驹,樊少冰,权磊..基于不确定性量化理论的寒区道路服役性能预测及评价研究[J].中外公路,2025,45(1):60-66,7.基金项目
国家重点研发计划项目(编号:2023YFB2604800) （编号:2023YFB2604800）
