基于数据融合的公交客流规模测算方法  

Passenger Volume Estimation Based on Data Fusion

作　　者：李林波[1] 姜屿[1] 王婧[2] 吴兵[1] 

机构地区：[1]同济大学道路与交通工程教育部重点实验室,上海201804 [2]济南市规划设计研究院,山东济南250101

出　　处：《城市交通》2016年第1期43-50,共8页Urban Transport of China

基　　金：国家自然科学基金面上项目"基于出行服务链的城镇群交通模式发展研究"(51178346)

摘　　要：公交客流规模测算往往存在调查成本受限和准确度要求较高的矛盾。提出基于公交IC卡历史数据与人工补充调查数据的数据融合测算方法,以准确推算公交客流规模。首先根据公交线路的基本属性,采用聚类分析方法划分线路类型,从每一类中选择具有代表性的线路。基于IC卡数据分析公交客流时变特征,运用有序样本聚类Fisher算法将线路小时刷卡量进行聚类分析。划分刷卡量相似时段,进而采用优化方法确定调查抽样率,确定相应的调查车辆进行人工补充调查,最终经过数据融合计算获得公交客流规模。基于上海市某辖区IC卡数据进行案例分析,测算得到三类公交线路的日均客流量。Accurate passenger flow estimation through surveys does not come without costs. This paper proposes a data fusion method based on the data from public transit IC card and supplementary surveys to accurately estimate passenger flow information. This paper first divides bus service routes into groups by their characteristics using cluster analysis method, and then selects one representative route from each group. Based on the temporary variation of bus passenger flow extracted from IC cards data, the paper categorizes IC card charging records per hour using Fisher algorithm of ordered sample cluster. By grouping time periods with similar IC card charging volumes, the paper determines the optimized sample rate and corresponding buses for the supplementary surveys. Consequently, bus passenger flows are estimated by data fusion method. Taking one district in Shanghai as an example, the paper demonstrates how to estimates daily passenger volumes of three types of bus routes using the above method.

关 键 词：公共交通 客流规模 聚类分析 数据融合 公交IC卡 上海市 

分 类 号：U491.17[交通运输工程—交通运输规划与管理]