基于ARIMA和GM模型的沥青路面使用性能预测  

Prediction of Service Performance of Asphalt Pavement Based on ARIMA and GM Models

作　　者：陈凯 黄丹 李瑶 CHEN Kai;HUANG Dan;LI Yao(School of Civil Engineering and Architecture,Wuhu Institute of Technology,Wuhu Anhui 241003)

机构地区：[1]芜湖职业技术学院建筑工程学院,安徽芜湖241003

出　　处：《湖北理工学院学报》2024年第2期45-49,共5页Journal of Hubei Polytechnic University

基　　金：安徽省高校科学研究项目(项目编号:KJ2021A1321);芜湖职业技术学院自然科学研究项目(项目编号:wzyzrzd202102,wzyrwzd202106)。

摘　　要：根据高速公路沥青路面使用性能的衰变规律可以确定养护维修道路的最佳时机。文章结合某高速公路沥青路面的技术状况(PQI),建立考虑时间趋势项的ARIMA模型和GM(1,1)模型,根据信息准则选取最优ARIMA预测模型,通过相对误差对两种模型进行权重优化,获得ARIMA-GM(1,1)组合模型。研究表明,3种路面使用性能预测模型都有良好的预测效果,其中GM(1,1)模型预测的相对误差在0.07%~1.18%之间;考虑时间趋势项的ARIMA模型的相对误差在0.07%~0.75%之间;组合模型的相对误差在0.07%~1.01%之间。在2010—2014年PQI数据预测中,ARIMA模型、GM(1,1)模型和组合模型的平均相对误差分别为0.266%、0.252%、0.246%,组合预测模型表现出了一定的优势,其预测结果可为养护计划决策提供一定的科学依据。The optimal timing of maintenance and repair of roads can be determined based on the decay laws of highway asphalt pavement serviceability.Combining the technical condition of asphalt pavement(PQI)of a highway,this paper established the ARIMA model and GM(1,1)model considering the time trend term,selected the optimal ARIMA prediction model according to the information criterion,and optimized the weights of the two models through the relative error to obtain the combined ARIMA-GM(1,1)model.The results showed that the three prediction models had good prediction effects,and the relative error of the GM(1,1)model prediction was between 0.07%and 1.18%;the relative error of the ARIMA model considering the time trend term was between 0.07%and 0.75%;the relative error of the combined model was between 0.07%and 1.01%.In the prediction of PQI data from 2010 to 2014,the average relative errors of ARIMA model,GM(1,1)model and combined model were 0.266%,0.252%and 0.246%respectively,and the combined prediction model showed certain advantages,and its prediction results can provide certain scientific basis for decision-making of the maintenance plan.

关 键 词：路面使用性能 ARIMA模型 GM(1 1)模型 组合模型 

分 类 号：U416.217[交通运输工程—道路与铁道工程]