基于轨迹数据的城市交通需求热点区域推荐研究

发布时间：2019-05-25 03:06

【摘要】：现阶段城镇化进程的不断加快、人们生活水平日益提高导致私家车数量的急剧增长,城市道路交通压力不断增大,急需对产生的大量交通数据进行分析来指导市政道路规划和提高城市管理水平,从繁杂的交通数据中发现城市运行的规律。同时随着无线通信技术和智能移动终端的高速发展,移动对象的轨迹数据采集变得更加便捷。出租车轨迹数据以其易收集、分布广和数据量大的特点,使得对出租车GPS轨迹数据进行数据挖掘成为近几年研究的热点。当前出租车行业存在空载率过高及打车难等问题,因此为出租车用户提供载客推荐服务具有重要研究意义。本文通过研究大量的出租车GPS数据,对出租车载客停留点进行了分析,利用数据挖掘技术发现轨迹数据中隐藏规律,对出租车载客热点区域及推荐方法进行了深入研究,为出租车用户提供载客热点推荐服务。本文研究的目的是最大限度的减少出租车空载巡游、减少城市交通污染、缓解交通压力并为出租车运营管理提供有价值的参考。首先,对出租车GPS数据及路网数据进行预处理,用于后期的数据分析与处理。使用ArcGIS平台对路网进行二次编辑,对获取的路网数据进行拓扑处理、完善属性字段并建立网路数据集加以验证。采用适合于低频采样的地图匹配方法,加入路网拓扑和速度约束条件,将车辆的位置信息与电子地图上的路网信息进行校对确定车辆在道路网中的实际位置。其次,利用统计分析方法,获得了出租车载客停留点随时间变化的规律,提出了一种基于粒子群的半监督近邻传播算法(PSO-SAP),用于出租车候选载客热点的发现,并在ArcGIS平台上集成PSO-SAP算法,对轨迹数据进行时空分析,结合ArcGIS平台实现热点区域展示,分析出出租车不同时间段载客热点区域的分布情况。最后,提出了结合出租车用户的信任度为出租车用户推荐载客概率大的热点区域,实验表明推荐结果具有较高的准确率。
[Abstract]:At present, with the continuous acceleration of urbanization, the increasing living standards of people lead to the rapid growth of the number of private cars, and the pressure of urban road traffic is increasing. It is urgent to analyze a large number of traffic data to guide municipal road planning and improve the level of urban management, and to find the law of urban operation from the complicated traffic data. At the same time, with the rapid development of wireless communication technology and intelligent mobile terminal, the trajectory data acquisition of mobile objects becomes more convenient. Taxi trajectory data has become a hot research topic in recent years because of its easy collection, wide distribution and large amount of data, which makes the data mining of taxi GPS trajectory data become a hot research topic in recent years. At present, there are some problems in taxi industry, such as high no-load rate and difficult taxi ride, so it is of great significance to provide passenger recommendation service for taxi users. In this paper, through the study of a large number of taxi GPS data, the taxi passenger stop point is analyzed, the hidden law in the track data is found by using data mining technology, and the hot spot area and recommendation method of taxi passenger are deeply studied. Provide hot spot recommendation service for taxi users. The purpose of this paper is to reduce taxi no-load cruise, reduce urban traffic pollution, relieve traffic pressure and provide valuable reference for taxi operation and management. First of all, the taxi GPS data and road network data are preprocessed for later data analysis and processing. The network is edited twice on ArcGIS platform, the data of road network is topologically processed, the attribute field is perfected and the network data set is established to verify it. The map matching method suitable for low frequency sampling is adopted, and the road network topology and speed constraints are added to calibrate the vehicle position information with the road network information on the electronic map to determine the actual position of the vehicle in the road network. Secondly, by using the statistical analysis method, the variation of taxi passenger stop with time is obtained, and a semi-supervised nearest neighbor propagation algorithm based on particle swarm optimization (PSO-SAP) is proposed, which can be used to discover the hot spots of taxi candidates. The PSO-SAP algorithm is integrated on ArcGIS platform, and the trajectory data are analyzed in time and space. Combined with ArcGIS platform, the hot spot area display is realized, and the distribution of taxi passenger hot spot area in different time periods is analyzed. Finally, a hot area with high probability of taxi user recommendation is proposed, which combines the trust degree of taxi users. The experimental results show that the recommendation results have high accuracy.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13;U491

【参考文献】