基于混合效应模型的沥青路面使用性能预测  

Prediction of Asphalt Pavement Performance Based on Mixed Effect Model

作　　者：杨国峰[1] 王浩仰 潘玉利[2,3,4] YANG Guo-feng;WANG Hao-yang;PAN Yu-li(School of Transportation,Southeast University,Nanjing Jiangsu 210096,China;Research Institute of Highway,Ministry of Transport,Beijing 100088,China;Roadmaint Co.,Ltd.,Beijing 100095,China;National Engineering Research Center of Road Maintenance Technologies,Beijing 100095,China)

机构地区：[1]东南大学交通学院,江苏南京210096 [2]交通运输部公路科学研究院,北京100088 [3]中公高科养护科技股份有限公司,北京100095 [4]公路养护技术国家工程研究中心,北京100095

出　　处：《公路交通科技》2018年第8期19-27,共9页Journal of Highway and Transportation Research and Development

基　　金：北京市交通行业科技项目(kj2015-4)

摘　　要：以华北某地区普通公路干线网为研究对象,根据2008—2015年路面损坏指数PCI、路面行驶质量指数RQI以及路面车辙指数RDI实测数据,结合路面结构、交通量、养护历史和地形等不同因素,利用统计分析技术建立和标定沥青路面使用性能预测模型。首先选定模型的预测指标及其影响变量,同时对相关数据资料进行收集和预处理,针对数据具体特点,以分析标定法为研究手段,采用混合效应模型方法对模型中的参数进行估计,最终通过实例验证模型的精度。结果表明:(1)PCI,RQI和RDI预测的平均误差分别为1.8,0.5和0.8。残差分布接近正态,说明模型很好地解释了数据的变异性;(2)模型简单,预测精度高,并且解释和量化了时间、养护类型、地形、行政等级、养管单位、技术等级、路面厚度以及交通量等因素对指标预测的影响。Based on the general trunk road network of a certain area in North China, according to the measured data of Pavement Condition Index （PCI）, Riding Quality Index （RQI） and Rutting Depth Index （RDI） during 2008--2015, combining with different factors such as pavement structure, traffic volume, maintenance history and topography, by using statistical analysis technology, the asphalt pavement performance prediction model is established and calibrated. First, the prediction index and the influencing variable of the model are selected. Then, the related data are collected and preprocessed, and the parameters of the model are estimated by means of analysis and calibration method and the mixed effect model approach according to the specific characteristics of the data. Finally, the accuracy of the model is verified through the examples. The result shows that （ 1 ） the average errors of PCI, RQI and RDI are 1.8, 0. 5 and 0. 8 respectively, the residual error distribution is close to normal, indicating that the model explains the variability of the data very well; （2） the model is simple and the prediction precision is satisfactory, it also explained and quantified the impact of time, maintenance type, topography, administrative level, maintenance unit, technology level, pavement thickness and traffic volume on the prediction of indicators.

关 键 词：道路工程 路面使用性能 分析标定 混合效应模型 预测模型 

分 类 号：U418.1[交通运输工程—道路与铁道工程]