基于手机定位数据的居民出行OD矩阵获取方法研究

发布时间：2018-02-12 10:44

本文关键词： OD调查 OD矩阵手机定位停留点出行链交通小区　出处：《西南交通大学》2016年硕士论文　论文类型：学位论文

【摘要】：城市交通是城市发展的基础和前提,是联系当代社会经济活动的纽带,若是想要使一个城市快速发展起来,那么我们必须首先解决其交通问题。解决交通问题主要包括交通问题的诊断、交通系统的规划和交通管理政策的制定,其一个重要的参考依据便是居民出行OD调查(Origin Destination Survey)。居民出行OD调查即调查居民出行从出发地点到目的地点的全面情况,了解居民出行的构成、流量以及去向。传统的居民出行OD调查方法主要采取人工询问和表格调查的方式,如路边询问法、表格调查法、电话调查法等,这些传统调查方法需要耗费大量的人力、财力,并且其抽样率较低、得到的OD矩阵精确度不高：基于现代先进技术的出行OD调查方法主要包括微波检测法、地感线圈检测法、视频图像检测法、GPS浮动车等,这些方法存在需要改动现有设备、安装维护费用高、覆盖面较小等不足。随着移动通信技术的迅速发展,手机普及率逐渐升高,大量的手机用户在日常出行中产生了海量的定位数据,这些定位数据的采集无需更改现有的设备,并且具有全天候连续采集、抽样率高、数据量庞大的特点,本论文基于手机定位数据的诸多优点研究了利用这些手机定位数据来获取居民出行OD矩阵的方法。本文首先简述了论文研究的背景与意义、国内外研究现状以及本文的主要研究内容,并就本研究相关的理论基础进行了介绍；分析了手机定位数据中主要存在的问题,并针对其存在的数据缺陷、定位漂移、定位数据非等时采集等问题进行了相应的数据预处理；然后提出了一种基于“空间-时间”双层聚类的停留点识别算法,先在空间层面利用DBSCAN算法对定位点进行聚类,并利用K-近邻法的思想优化聚类结果,得到候选停留点,再在时间层上对定位数据聚类以对候选停留点进行划分和取舍,识别出用户出行的停留点和出行链；考虑到聚类和交通小区划分之间存在很大的相似性,利用K-Means算法对停留点聚类实现交通小区划分,其中,K-Means算法中的参数k根据交通小区划分原则确定：然后将出行链映射到划分好的交通小区中,从中识别区外出行,再经过统计计算后得到居民出行OD矩阵：最后基于潍坊市电信用户手机定位数据,设计实验验证了基于手机定位数据的居民出行OD矩阵获取方法,并结合GIS分析了潍坊市OD客流的地理空间分布特征。
[Abstract]:Urban transportation is the foundation and premise of urban development and the link to contemporary social and economic activities. If you want to make a city develop rapidly, Well, we must first solve its traffic problems. Solving traffic problems mainly includes the diagnosis of traffic problems, the planning of traffic systems and the formulation of traffic management policies. One of its important reference bases is Origin Destination Survey. The residents' travel OD survey is to investigate the overall situation of residents' travel from departure place to destination, and to understand the composition of residents' travel. Flow and destination. Traditional OD survey methods for residents' travel mainly adopt manual inquiry and form investigation, such as roadside inquiry, tabular investigation, telephone survey, etc. These traditional survey methods require a large amount of manpower and financial resources. And its sampling rate is low and the precision of OD matrix is not high. The methods of OD survey based on modern advanced technology mainly include microwave detection method, ground inductance coil detection method, video image detection method and GPS floating vehicle, etc. With the rapid development of mobile communication technology, the penetration rate of mobile phones has gradually increased, with the rapid development of mobile communication technology, such as the need to modify existing equipment, the high cost of installation and maintenance, and the small coverage. A large number of mobile phone users have generated a large amount of location data in their daily travel. The acquisition of these location data does not need to change the existing equipment, and it has the characteristics of continuous acquisition, high sampling rate and huge amount of data. Based on the advantages of mobile phone location data, this paper studies how to use these mobile phone location data to obtain the OD matrix of resident travel. Firstly, the background and significance of this paper are briefly described. Domestic and foreign research status and the main research content of this paper, and the related theoretical basis of this study are introduced, the main problems in mobile phone positioning data are analyzed, and in view of its data defects, positioning drift, In this paper, the corresponding data preprocessing is carried out for the non-isochronous data acquisition, and then an algorithm based on "space-time" two-layer clustering is proposed to identify the location points. Firstly, the DBSCAN algorithm is used to cluster the location points at the spatial level. Using the idea of K- nearest neighbor method to optimize the clustering results, the candidate stopover points are obtained, and then the location data are clustered on the time level to divide and choose the candidate stopover points, and the user travel stopover points and trip chains are identified. Considering that there is a great similarity between clustering and traffic cell division, K-Means algorithm is used to realize traffic cell division. The parameter k of K-Means algorithm is determined according to the principle of traffic cell division. Then the trip chain is mapped to the divided traffic area, from which the travel outside the area can be identified. Finally, based on the mobile phone location data of telecom users in Weifang, the method of obtaining OD matrix based on mobile phone positioning data is designed and verified. Combined with GIS, the spatial distribution characteristics of OD passenger flow in Weifang are analyzed.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：U491.12;TP311.13

【相似文献】