基于GPS数据的公交站点区间行程时间分布与可靠性分析

发布时间：2018-05-05 01:25

本文选题：公交 + 行程时间可靠性　；参考：《浙江大学》2015年硕士论文

【摘要】：公交行程时间可靠性是城市公共交通系统运行状态评价的重要指标之一,能够准确表征公交行程时间的稳定性,为公交出行者、公交管理部门和交通专家提供评价公交服务水平的重要依据。同时,移动检测技术的日益发展为采集数据的实时性、动态性、准确性提供了保障。本文在对公交GPS数据进行分析、挖掘的基础上,建立了公交站点区间行程时间可靠性评价方法,为公交运营状态评价提供了一种新思路。首先,本文对公交GPS数据和GIS数据从数据结构、数据内容、数据保存三方面做了简要概述,并在传统的地图匹配算法的基础之上,提出了改进的投影匹配算法,采用了网格分块的快速匹配方法实现公交GPS数据的匹配。针对GPS漂移数据,利用匹配阈值法对其判断并予以剔除；并建立了完全重复冗余数据的处理方法。实验证明,上述GPS数据预处理方法具有良好的效果,能够用于公交行程时间的计算。接着明确了站点区间行程时间的定义,采用4类不同形式的分布模型：正态分布、对数正态分布、威布尔分布、高斯混合分布分别对高峰、平峰时段的公交行程时间数据进行拟合优度检验,结果表明高斯混合分布模型能够最好地拟合公交行程时间数据。通过EM算法求解高斯混合模型的各项参数,并建立了站点区间行程时间分布的拟合步骤。把现有的行程时问可靠性指标归结为三类：基于概率的评价指标、基于偏差的评价指标和介于两者之间的评价指标,并在此基础上,结合站点区间行程时间的三元高斯混合分布模型,采用波动性指数和延误指数作为表征行程时间可靠性的指标,建立了基于高斯混合分布模型的站点区间行程时间可靠性评价方法。最后,分析了影响公交站点行程时间可靠性的几大因素：道路几何条件(站点区间长度和交叉口相对位置)、交通状态(公交车小时流量因素、交叉口信号因素)、天气因素和采样间隔。上述研究将为公交运营管理、信息发布、服务水平评价等提供技术支撑。
[Abstract]:The reliability of bus travel time is one of the important indexes to evaluate the running state of urban public transportation system. It can accurately represent the stability of bus travel time and is a public transport traveler. Public transport management departments and traffic experts provide an important basis for evaluating the level of public transport services. At the same time, the development of mobile detection technology provides a guarantee for the real-time, dynamic and accuracy of data collection. In this paper, based on the analysis and mining of bus GPS data, the reliability evaluation method of bus stop interval travel time is established, which provides a new way for bus operation state evaluation. First of all, this paper gives a brief overview of bus GPS data and GIS data from three aspects: data structure, data content and data preservation. Based on the traditional map matching algorithm, an improved projection matching algorithm is proposed. The fast matching method of grid block is used to realize the matching of bus GPS data. According to the GPS drift data, the matching threshold method is used to judge and eliminate the data, and a method to deal with the redundant data is established. The experimental results show that the GPS data preprocessing method is effective and can be used to calculate the bus travel time. Then the definition of travel time between stations is defined, and four kinds of distribution models are adopted: normal distribution, logarithmic normal distribution, Weibull distribution, Gao Si mixed distribution. The results show that the mixed distribution model of Gao Si can best fit the bus travel time data. The parameters of Gao Si's mixed model are solved by EM algorithm, and the fitting steps of travel time distribution between stations are established. The existing travel time reliability index is divided into three categories: probability-based evaluation index, deviation-based evaluation index and evaluation index between the two, and on this basis, Combined with the ternary Gao Si mixed distribution model of travel time between stations, volatility index and delay index are used as indicators to characterize the reliability of travel time. The reliability evaluation method of interval travel time based on Gao Si mixed distribution model is established. Finally, several factors affecting the reliability of bus stop travel time are analyzed: road geometry conditions (station interval length and intersection relative position, traffic state (bus hour flow factor), Intersections signal factors, weather factors and sampling intervals. The above research will provide technical support for bus operation management, information release, service level evaluation and so on.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491

【相似文献】