基于LBS轨迹的出行活动链模式识别研究

发布时间：2018-05-30 04:09

  本文选题：城市交通 + 出行链　； 参考：《大连交通大学》2015年硕士论文

【摘要】：互联网技术的变革、移动通信技术的应用、智能交通技术的成熟,为传统的出行信息调查、出行行为研究、出行需求预测提供了新的思路。本文的研究目的,是希望建立一种方法,能够有效地从利用智能手机定位及相关位置信息所采集到的出行轨迹数据中,提取出出行方式、活动类型等信息,从而提升居民出行调查的效率、降低调查过程中的主观性、减少调查周期和费用,为城市交通规划与管理提供数据支撑和决策支持。本文的研究结合了出行链、模式识别、被动式居民出行信息调查和手机位置服务(LBS)等理论和技术基础。在研究活动链和出行链结构的基础上,建立了出行活动链模式,划分了出行过程子模式和活动过程子模式,并分析了其模式特征,研究了出行活动链模式和出行轨迹之间的对应关系;利用手机定位和传感器模块,结合基于LBS的丰富位置信息的采集思想,构建了出行轨迹数据的采集方法,并且应用了轨迹插值来补全轨迹中的缺失点,采用Kalman滤波来实现轨迹降噪,提出滑窗判别的方法将轨迹划分成出行段和活动段;建立了出行过程子模式和活动过程子模式的特征向量,并给出了从出行轨迹参数向量中提取子模式特征向量的方法,采用频率分布图和F-score的方法对特征向量在两两分类间的可分性进行了定性和定量的分析,进而采用了决策树、BP网络、RBF网络和支持向量机等分类器对样本数据进行识别。最后以大连市为背景实地采集了出行轨迹数据,并利用这些数据应用上述方法进行了实证研究,对于数据补全、滤波、分段、识别等方法的效果进行了评价,结果表明本文所应用的方法对于利用LBS轨迹来进行出行活动链模式识别能够取得较好的效果。
[Abstract]:The innovation of Internet technology, the application of mobile communication technology and the maturity of intelligent transportation technology provide a new way of thinking for traditional travel information investigation, travel behavior research and travel demand prediction. The purpose of this paper is to establish a method, which can extract the information of travel mode, activity type and so on from the travel path data collected by using the location information of smart phone and related location information. In order to improve the efficiency of residents' travel survey, reduce the subjectivity of the survey process, reduce the investigation cycle and costs, and provide data support and decision support for urban traffic planning and management. This paper combines the theory and technology of trip chain, pattern recognition, passive travel information survey and mobile location service (LBS). On the basis of studying the structure of activity chain and trip chain, this paper establishes the travel activity chain pattern, divides the travel process sub-pattern and the activity process sub-pattern, and analyzes its pattern characteristics. The corresponding relationship between trip activity chain mode and trip trajectory is studied, and the acquisition method of trip trajectory data is constructed by using mobile phone location and sensor module, combined with the idea of collecting abundant location information based on LBS. The path interpolation is used to compensate the missing points in the whole trajectory, and the Kalman filter is used to reduce the trajectory noise. A sliding window discriminating method is proposed to divide the trajectory into travel segment and active segment. The Eigenvectors of travel process subpattern and activity process subpattern are established, and the method of extracting subpattern eigenvector from trip path parameter vector is given. The qualitative and quantitative analysis of the separability of feature vectors between pairwise classification is carried out by means of frequency distribution map and F-score, and then the classifiers such as decision tree BP neural network and support vector machine are used to identify the sample data. Finally, taking Dalian as the background, we collect the travel track data in the field, and use these data to carry on the empirical research with the above methods, and evaluate the effect of the data complement, filtering, segmentation, recognition and so on. The results show that the method proposed in this paper can achieve good results for pattern recognition of trip activity chain using LBS locus.
【学位授予单位】：大连交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495

【参考文献】

中国期刊全文数据库 前2条

1 苟博;黄贤武;;支持向量机多类分类方法[J];数据采集与处理;2006年03期

2 隽志才;鲜于建川;;交通信息作用下的活动-出行决策行为研究[J];中国公路学报;2008年04期

中国重要会议论文全文数据库 前1条

1 杨兆升;王媛;;基于手机探测车的交通信息采集方法研究[A];第一届中国智能交通年会论文集[C];2005年

中国硕士学位论文全文数据库 前1条

1 贾扬洋;基于LBS的实时交通信息系统的设计和实现[D];河南大学;2009年

，

本文编号：1953849