一级干线公路的Panel Data限速模型  

Panel Data Speed Limit Model for First-class Arterial Highway

作　　者：王华荣[1,2] 孙小端[3] 聂百胜[1] 

机构地区：[1]中国矿业大学(北京)资源与安全工程学院,北京100083 [2]交通运输部公路科学研究院北京交科公路勘察设计研究院有限公司,北京100191 [3]北京工业大学交通工程北京市重点实验室,北京100124

出　　处：《公路交通科技》2012年第4期114-119,共6页Journal of Highway and Transportation Research and Development

基　　金：交通运输部西部交通建设科技项目(200731800027)

摘　　要：为使限速值的确定更为客观,避免以往仅以85%位速度作为限速主要依据所存在的缺陷,在剖析合理限速与运行速度、道路特征等约束条件之间所存在逻辑关系的基础上,根据百分位速度与其他限速影响因素所构成变量的数据结构与Panel Data结构的相似性,引入Panel Data相关建模方法来构建限速与百分位速度等影响因素之间的理论模型。采用逐步回归法与最小二乘虚拟变量相结合的方法来求解最优时点固定效应模型表达式,应用F检验或Hausman检验结果确定Panel Data模型类型及具体表达式,最终确定一级干线公路小型车限速的主要影响因素依次为行人干扰、地形、纵坡、百分位速度,大型车限速的主要影响因素依次为地形、纵坡、行人干扰、出入口密度。最后,应用统计软件分别对一级干线公路大、小型车Panel Data限速模型残差有效性进行检验,结果表明所建模型误差均在±10 km/h之内,说明所提出的限速确定方法在实际应用中具备一定的可行性。In order to determine the speed limit value more objective and avoid the defects of setting speed limits at 85 percentile speeds, based on analyzing the logic relationship among constraint conditions, including reasonable speed limit, operating speed, road characteristic, etc. Then panel data method was introduced to build up the theoretical model of speed limit and influencing factors such as percentile speed, etc. due to its similarity with the data structure constituted by variables of percentile speed and other influencing factors of speed limit. The methods of stepwise regression and least square dummy variable （LSDV） were used to find the expression of optimum fixed time effect model, then F-test or Hausman test was used to determine the panel data model type and its specific expression. The results reveal that the main influencing factors of setting speed limits for small vehicles are pedestrian intervention, terrain, slope and percentile speed successively, while the main influencing factors of speed limit for large vehicle are terrain, slope, pedestrian intervention and density of access way successively. Finally, the effectiveness of the residuals of panel data speed limit models of first-class arterial highway for small and large vehicles was tested by statistical software. The result shows that the errors of the established models are within±10 km/h, which means the presented method for setting speed limit is feasible when applied in reality.

关 键 词：交通工程 一级干线公路 PANEL Data线性回归模型 限速 

分 类 号：U412.363[交通运输工程—道路与铁道工程]