公路轴载谱与当量轴次的深度学习计算及应用  

Deep Learning Calculation and Application of Highway Axle Load Spectrum and Equivalent Axle Volumes

作　　者：张路凯 撒蕾[1] 王英平[1] 孙满 刘坤 米雪玉 ZHANG Lu-kai;SA Lei;WANG Ying-ping;SUN Man;LIU Kun;MI Xue-yu(Transport Planning and Research Institute,Ministry of Transport,Beijing 100029,China;Xuzhou Highway Development Center,Xuzhou 221006,China;North China University of Technology,Tangshan 063210,China)

机构地区：[1]交通运输部规划研究院,北京市100029 [2]徐州市公路事业发展中心,徐州市221006 [3]华北理工大学,唐山市063210

出　　处：《公路》2023年第12期1-6,共6页Highway

基　　金：国家科技创新2030重大项目,项目编号2022ZD0115600;国家重点研发计划项目,项目编号2022YFC3703600。

摘　　要：以高速公路与普通国省道为对象,旨在实现全网范围的轴载谱和当量轴次计算。首先,以公路沥青路面设计规范为依据,分析路段轴载谱和累计当量轴次的计算需求;之后,针对现有轴载检测覆盖条件,提出以公路交通情况观测调查数据为支撑的计算方法;进而,针对多源数据的融合目标,建立广义回归神经网络模型,实施深度学习计算流程;最后,以徐州市公路网为实例,对所提出方法进行验证应用。推算结果中,全网路段平均轴载谱与实际规律总体一致;验证样本中3轴车偏差最小,6轴车偏差最大;通过计算当量轴次,获得了用于养护决策支持的区域重载路段分布和分级。The expressways,national highways and provincial roads are taken as study objects,with the aim to realize the calculation of the axle load spectrum and equivalent axle volumes of the whole network.Firstly,based on the Specifications for Design of Highway Asphalt Pavement,the calculation requirements of the axle load spectrum and the cumulative equivalent axle volume are analyzed.Then,according to the existing axle load detection coverage conditions,the calculation supported by the observation and survey data of highway traffic conditions is clarified.In accordance with the fusion objective of multi-source data,ageneralized regression neural network model is established to execute the deep learning calculation process.Finally,the highway network of Xuzhou is taken as the background to verify the application of the proposed method.In the calculation results,the average axle load spectrum of the whole network is generally consistent with the actual rules.The deviation of the 3-axle vehicle in the verification sample is the smallest,and the deviation of the 6-axle vehicle is the largest.By calculating the equivalent axle volume,the regional heavy-duty road distribution and ranking for maintenance decision support are obtained.

关 键 词：公路工程 轴载谱 当量轴次 深度学习 

分 类 号：U491.113[交通运输工程—交通运输规划与管理] U416.2[交通运输工程—道路与铁道工程]