河北省科学院学报

从AFC数据中挖掘出规律出行的乘客，并对该类乘客在早晚高峰出行的地铁站点进行分类，对交通管理部门引导乘客出行与缓解车厢和站点拥挤具有重要意义。本研究以AFC数据为基础，提出了一种基于SA-PSO聚类算法的乘客出行规律挖掘方法。利用该方法基于广州地铁的AFC数据对乘客出行特征进行聚类分析，将城市地铁乘客主要分为2类，规律出行乘客（占比53.21%）和非规律出行乘客（占比46.79%）；同时，基于城市POI数据及规律出行乘客的早晚高峰出行站点客流进行分析。结果表明，地铁站点可分为3类，即工作密集型、居住密集型及职住混合型。该研究可为交通管理部门制定更加精准的乘客出行引导策略及优化站内配置提供新的思路。

Identifying passengers with regular travel patterns from Automated Fare Collection(AFC)data and classifying their frequently used subway stations during morning and evening peaks helps transportation management departments guide passenger flow and alleviate congestion in carriages and stations.Based on AFC data,this study proposes a method for mining passenger travel regularity using a Simulated Annealing-Particle Swarm Optimization(SA-PSO)clustering algorithm.By applying this method to cluster and analyze passenger travel characteristics based on Guangzhou Metro′s AFC data,urban subway passengers are primarily categorized into two groups:regular passengers(53.21%)and nonregular passengers(46.79%).Simultaneously,based on urban Point of Interest(POI)data and passenger flow at stations used by regular passengers during morning and evening peaks,subway stations can be classified into three types:work-intensive,residence-intensive,and job-housing mixed.This research can provide new insights for transportation management departments to formulate more precise passenger flow guidance strategies and optimize station configurations.