082.基于智能交通卡数据的轨道出行乘客特征研究.pdfVIP

基于智能交通卡数据的轨道出行乘客特征研究 曾志南 摘要：近年来国内大中城市轨道交通快速发展，各大城市轨道交通网络初步形成，客流网络化特征的日趋 明显，乘客的路径选择行为存在差异，网络客流时空分布不均衡。研究轨道出行乘客出行规律，有助于理 解城市轨道交通系统与职住空间的耦合机制，为城市轨道交通体系的优化和调整提供理论支持和定量依据。 本文对于轨道出行乘客特征研究提供一种新的思路，基于深圳市2015年12月的智能交通卡数据，首 先对智能交通卡数据进行数据清洗，其次以一个月全天、一个月的工作日、一个月的工作日早高峰、一个 月的周末四个时间段为限定条件，分别提取每位用户的特征值，同时引入了“信息熵”的概念来分别计算 四个时间段乘客出行的 “纯度”。在此基础上使用k-means聚类算法首先将5329267位乘客分为各有特征 的14 小类。通过进一步描述小类特征，将14 小类出行特征总结为常规出行特征、半常规出行特征、随机 出行特征三大类。研究分析了三种不同特征乘客对于站点的使用，并结合乘客出行特征，进一步分析了各 出行特征的主体人群。最后，研究也注意到站点周边土地利用情况与站点的不同特征乘客使用比例的关系， 提出了未来的研究方向。 关键词：轨道交通，智能交通卡数据，出行特征，k-means聚类算法，信息熵，深圳 Abstract：In recent years,railtransit in large andmedium citieshasbeen developingrapidly.Theurban railtransit network has initially formed, and the characteristics of passenger flow network become more and more obvious.Passengersroute choicebehavior is different, and the network passenger flow isnot balanced in time and space. Studying the law of passengers travelling on rail travel will help to understand the coupling mechanism between urban rail transit system andjobs and housing spatial, and provide theoretical support and quantitative basisfortheoptimization andadjustment ofurban railtransit system. This paper provides a new idea for the study of the characteristics of rail passengers, using the intelligent trafficcard dataof Shenzhen inDecember 2015. First cleaning the data. Secondly, the characteristic values of each user are extracted from the full day of one month, one months working day, one months work day peak, and four time periods of one months weekend.At the same time, the concept of information entropy is introduced to calculate the purity of passengers in four periods. On this basis, the K-means clustering algorithm is used to divide the 5329267 passengers into 14 small groupswith different characteristics.The characteristics of small classes are further described,and three categories ofregular trip characteris