基于手机定位数据的城市居民出行特征提取方法研究

发布时间：2018-01-08 14:07

本文关键词：基于手机定位数据的城市居民出行特征提取方法研究　出处：《东南大学》2015年硕士论文　论文类型：学位论文

【摘要】：利用手机定位技术采集居民出行信息具有实时性、低成本、样本大、易实施等优点,与传统出行调查优势互补,成为出行特征研究的又一重要数据来源。然而手机定位数据时间分布不均匀,且定位精度具有不确定性,影响因素繁复,目前已有的出行特征提取方法还存在一定不足。本研究基于大规模真实手机信令定位数据,从数据时空特性角度,尝试建立一种更为准确可行的出行特征提取方法,具有重要的理论与工程实践意义。第一,为了对定位数据产生的机制有一个全面深入的了解,研究对蜂窝通信网络和手机定位技术的基本原理进行介绍,并对手机信令数据内涵进行深入解读。在此基础上,结合出行的一般定义和手机定位数据特点,对一次“手机出行”进行定义,并进一步说明利用手机定位技术提取出行特征的适用性,为后续研究打下铺垫。第二,数据处理与特性分析。研究深入分析定位数据产生机制,并借鉴已有的研究成果,建立一种更为完善的多层次数据处理方法,包括对多种无效数据过滤、多种噪音数据识别处理等,进而为出行特征提取提供了高质量数据来源。然后,研究从事件类型、时间、距离、平均速度等多个维度,分析数据特性,为建立出行特征提取模型提供依据。第三,建立出行特征提取模型。在以上研究的基础上,研究深入分析手机定位轨迹点时空特征,发现手机用户出行轨迹点主要由代表停留的圆形停留区域以及代表出行的长条形出行区域组成。基于此,研究提出一种时空聚类算法识别圆形停留区域,将轨迹点划分为停留点和运动点,同时获取停留起讫时间、停留位置等多种出行信息。基于以上信息,研究建立了出行次数、出行距离、出行速度、出行时间分布等多种出行特征指标的计算方法。最后,研究多角度分析了从手机出行到用户出行的扩样影响因素,建立多层扩样方法。第四,实例分析与验证。首先,研究结合“手机出行”定义,对模型中每个参数进行影响分析和标定。然后,研究选取上海市某工作日大规模手机定位数据进行示例分析,并将模型分析结果与上海市第四次出行调查数据进行对比分析。对比发现出行次数、出行时间分布等多个出行特征指标与居民调查数据高度相关,CORREL检验值在0.92以上,说明研究提出的出行特征提取方法具有较好的可靠性和适用性。最后,研究总结模型方法的创新与不足,并对后续研究提出展望。
[Abstract]:The use of mobile phone positioning technology to collect resident travel information has the advantages of real-time, low cost, large sample, easy to implement, and complementary with the traditional travel survey. It has become another important data source for the study of travel characteristics. However, the time distribution of mobile phone location data is uneven, the location accuracy is uncertain, and the influencing factors are complicated. At present, there are still some shortcomings in the existing travel feature extraction methods. This study is based on large-scale real mobile phone signaling location data, from the point of view of data space-time characteristics. Try to establish a more accurate and feasible travel feature extraction method, which has important theoretical and engineering significance. First, in order to have a comprehensive and in-depth understanding of the mechanism of location data generation. This paper introduces the basic principles of cellular communication network and mobile phone location technology, and deeply interprets the meaning of mobile phone signaling data. On this basis, combined with the general definition of travel and mobile phone location data characteristics. To define a "mobile travel", and further explain the use of mobile phone location technology to extract travel characteristics of applicability, for the follow-up study lay the groundwork. Second. Data processing and characteristic analysis. Research on location data generation mechanism, and use for reference of existing research results, establish a more perfect multi-level data processing method, including a variety of invalid data filtering. A variety of noise data processing, which provides a high-quality data source for travel feature extraction. Then, from the event type, time, distance, average speed and other dimensions, analysis of data characteristics. In order to establish a travel feature extraction model to provide the basis. Third, establish a trip feature extraction model. On the basis of the above research, in-depth analysis of mobile phone locus locus space-time features. It is found that the mobile phone user travel path points are mainly composed of the circular stay area which represents the stay and the long travel area which represents the trip. Based on this, a spatio-temporal clustering algorithm is proposed to identify the circular stay area. Track points are divided into stopover points and motion points, and a variety of travel information, such as stop time, stop position and so on, are obtained. Based on the above information, the number of trips, travel distance and travel speed are studied. Finally, the paper analyzes the influence factors of mobile phone travel to user travel, and establishes a multi-layer sample expansion method. 4th. First, combining the definition of "mobile phone trip", each parameter in the model is analyzed and calibrated. The study selected a large-scale mobile phone location data for a workday in Shanghai for example analysis, and the results of model analysis and Shanghai 4th trip survey data were compared and analyzed. Travel time distribution and other travel characteristics are highly correlated with resident survey data and Correl test value is more than 0.92. The results show that the proposed method has good reliability and applicability. Finally, the innovation and deficiency of the model method are summarized, and the prospect of the future research is put forward.
【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491

【相似文献】