铁道运输与经济2025，Vol.47Issue(7)：44-55,12.DOI:10.16668/j.cnki.issn.1003-1421.2025.07.05
城市轨道交通短时客流预测研究综述及展望
Review and Prospects of Short-Term Passenger Flow Prediction in Urban Rail Transit
摘要
Abstract
Short-term passenger flow prediction in urban rail transit is essential for dynamically optimizing capacity allocation and intelligent station management.A scenario-based review is of great significance for establishing a multi-scenario prediction system.Based on normal,predictable,and unpredictable scenarios,research progresses on station-level passenger flow(inbound/outbound)and origin-destination(OD)passenger flow prediction were summarized in terms of research question,feature engineering,and method model.According to the findings,research for normal scenarios is relatively mature,though real-time model parameter adaptation needs further improvement.Predictable and unpredictable scenarios present significant challenges brought by incomplete mechanistic understanding and insufficient data.The unpredictable scenarios are especially severe due to their burstiness and absence of historical patterns.Therefore,predictable scenarios require a deeper analysis of event-specific characteristics and integration of techniques like transfer learning to alleviate data scarcity.Unpredictable scenarios need to elucidate the link between micro-level passenger behavior and macro-level flow patterns.The scenario coverage should be expanded through data driving to improve pattern libraries.The robustness and generalizability of the model are thus enhanced.关键词
城市轨道交通/短时客流预测/常态场景/不可预知场景/可预知场景Key words
Urban Rail Transit/Short-Term Passenger Flow Prediction/Normal Scenario/Unpredictable Scenario/Predictable Scenario分类
交通工程引用本文复制引用
凌晨,许心越,叶子扬,金灏涵,宋佳玺..城市轨道交通短时客流预测研究综述及展望[J].铁道运输与经济,2025,47(7):44-55,12.基金项目
国家重点研发计划项目(2022YFC3005204) （2022YFC3005204）
