基于大规模车辆轨迹数据的道路交叉口排队长度探测

发布时间：2018-07-24 08:37

【摘要】：道路交叉口排队长度的探测能够对交通拥堵形成有效的监测与预防,对缓解路网压力、促进城市交通健康发展有着积极的作用。浮动车运行时所产生的轨迹数据(简称浮动车GPS数据)是一种典型的移动式数据,针对这些数据进行分析与挖掘,从而实现排队长度的探测是当下的研究热点。然而,现有研究往往存在工程应用条件严苛和参数设定只能针对具体道路交叉口的适用性问题,同时缺乏对数据样本容量的相关研究。针对这些问题,本文以深圳市大规模浮动车GPS数据为基础,提出一种基于空间分布模式的道路交叉口排队长度探测算法,并研究相应的最佳样本容量。论文首先对数据进行预处理,综合道路中介中心性和浮动车流量得到不同属性的道路集合,在分析原始数据采样间隔的分布后选取了统一的轨迹采样间隔;然后根据排队形成的机理,提取排队车辆的轨迹点特征,通过引入空间分布模式的分析方法,定量地获得了道路上轨迹点的空间点格局,并结合提取的特征实现了排队长度的探测;最后,分析浮动车GPS样本容量对探测结果和算法效率的影响,并利用核密度估计原理得到不同样本容量下的轨迹点在道路上的密度分布曲线,通过定量地度量曲线稳定性,确定最佳样本容量。实验结果表明,所提算法在保证高精度的前提下具有很强的适用性,可用于多种类型道路交叉口的排队长度探测。此外,本文确定的最佳样本容量具有很高的正确性,同时得出了属性相似道路形成的道路交叉口,排队长度探测的最佳样本容量具有一致性的结论。本文以浮动车GPS数据为研究对象,提出了具有强适应性的排队长度探测算法,并确定了相应的最佳样本容量,为大规模城市道路交叉口的车辆排队长度探测提供了新的方法与思路。
[Abstract]:The detection of queue length at intersections can effectively monitor and prevent traffic congestion, relieve road network pressure and promote the healthy development of urban traffic. The track data (GPS data for short) generated by floating vehicle is a kind of typical mobile data. It is a hot research topic to analyze and mine these data and realize the detection of queue length. However, the existing research often has the problem that the engineering application conditions are strict and the parameter setting can only be applied to the specific road intersection, and there is a lack of relevant research on the data sample size at the same time. In order to solve these problems, based on the GPS data of large-scale floating vehicles in Shenzhen City, this paper presents an algorithm of queue length detection for road intersections based on spatial distribution pattern, and studies the corresponding optimal sample size. Firstly, the data are preprocessed, and the road sets with different attributes are obtained by synthesizing the road intermediary centrality and floating vehicle flow. After analyzing the distribution of the sampling interval of the original data, the uniform track sampling interval is selected. Then according to the mechanism of queue formation, the characteristics of trajectory points of queued vehicles are extracted. By introducing the analysis method of spatial distribution mode, the spatial point pattern of trajectory points on the road is obtained quantitatively. Finally, the influence of the GPS sample size on the detection results and the efficiency of the algorithm is analyzed. Based on the kernel density estimation principle, the density distribution curves of trajectory points on the road under different sample sizes are obtained, and the optimal sample size is determined by quantitatively measuring the stability of the curves. The experimental results show that the proposed algorithm has strong applicability on the premise of high accuracy and can be used to detect the queue length of many kinds of road intersections. In addition, the optimal sample size determined in this paper has a high accuracy, and the conclusion that the optimal sample size of queue length detection is consistent is obtained for the road intersection with similar attributes. In this paper, based on the GPS data of floating vehicle, a strong adaptive queue length detection algorithm is proposed, and the corresponding optimal sample size is determined. It provides a new method and train of thought for vehicle queue length detection at large-scale urban road intersections.
【学位授予单位】：深圳大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491

【相似文献】