摘要
排队长度作为评价信号交叉口运行效率的一个重要指标,能有效反映交叉口处的运行状况.传统排队检测模型大多基于线圈检测器,且模型假设过于理想化,本文提出一种面向低采样率浮动车数据、具有良好数据驱动性的信控交叉口在线排队长度检测方法,方法关键在于利用队尾浮动车位置估算最大排队长度.检测过程采用固定时间间隔,主要步骤包括地图匹配、等距划分交叉口进口道并统计停车点数量、判定队尾浮动车的位置、修正得到最大排队长度估计值.实测数据表明,此方法的精度与浮动车比率有直接的关系,在浮动车比率较高的许多主干路交叉口,精度可以达到理想效果,30 m以内的平均绝对误差对高峰期的排队检测依旧具有很大价值.
Queue length is an important index to evaluate the operation efficiency of signalized intersections and reflect the traffic status of intersections.Conventional queue length estimation models are mainly based on loop detectors and idealistic assumptions.This paper proposes an online queue length estimation method at signalized intersections with low-sampling-rate floating car data.The core of the method is estimating the maximum queue length considering the position of the last floating car.Using fixed time intervals,the method is data-driven and consists of map matching,low-speed point statistics,determination of the last floating car and queuing correction.The ground test shows the precision of the method is closely related to the floating car rate.At trunk road intersections with high floating car rates,the expected accuracy can be obtained.For peak hours,the mean absolute error（MAE） within 30 meter is acceptable and of great value.
出处
《交通运输系统工程与信息》
EI
CSCD
北大核心
2013年第3期78-84,共7页
Journal of Transportation Systems Engineering and Information Technology
基金
国家高技术研究发展计划(863计划
2011AA110306)
关键词
智能交通
排队长度估计
浮动车
交叉口
低速点
intelligent transportation
queue length estimation
floating car
intersection
low-speed points
