基于NAR-ARIMA组合模型的高速公路沥青路面破损状况预测  

Prediction of expressway asphalt pavement damage based on NAR-ARIMA combined model

作　　者：李海莲[1] 高雅丽 江晶晶 司金忠 LI Hailian;GAO Yali;JIANG Jingjing;SI Jinzhong(School of Civil Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China;China Railway Liuyuan Group Co.,Ltd.,Tianjin 300308,China)

机构地区：[1]兰州交通大学土木工程学院,甘肃兰州730070 [2]中铁第六勘察设计院集团有限公司,天津300308

出　　处：《大连理工大学学报》2024年第3期307-313,共7页Journal of Dalian University of Technology

基　　金：国家自然科学基金资助项目(51868042);甘肃省自然科学基金资助项目(20JR10RA229,22JR5RA334);甘肃省高等学校创新基金资助项目(2021A-048);兰州交通大学“百名青年优秀人才培养计划”基金资助项目(2018103).

摘　　要：基于NAR神经网络模型和ARIMA传统时间序列预测模型,对高速公路沥青路面的破损状况进行预测,再分别通过最优加权法和残差优化法对两种预测模型进行组合,得到两种组合模型.对各单一模型和组合模型的精度和稳定性进行了比较分析.实例分析表明,组合模型相较于单一模型的精度和稳定性均有所提升,NAR-ARIMA最优加权组合模型预测效果最佳.该组合模型所需样本量较小,且基于时间序列.由于采用历史数据作为影响因素代替指标,该组合模型计算速度快、精度高,适用于日常的预测工作,为后续合理的道路养护决策提供了重要的理论依据.Based on NAR neural network model and ARIMA traditional time series prediction model,expressway asphalt pavement damage is predicted,then the two prediction models are combined by the optimal weighting method and the residual optimization method,respectively,two combined models are obtained.The accuracy and stability of each single model and combined model are compared and analyzed.The case analysis shows that the accuracy and stability of the combined model are improved compared with the single model,and the NAR-ARIMA optimal weighted combined model has the best prediction effect.This combined model requires a small sample size and is based on time series.Because historical data is used as influencing factors instead of indices,this combined model has fast calculation speed and high accuracy,which is suitable for daily prediction work and provides an important theoretical basis for subsequent reasonable pavement maintenance decisions.

关 键 词：道路工程 路面破损状况预测 ARIMA模型 NAR神经网络模型 沥青路面 

分 类 号：U416.217[交通运输工程—道路与铁道工程]