基于不同时间粒度的耦合时空特征下地铁短时客流预测  

Prediction of Subway Short-term Passenger Flow Based on Coupled Time-space Characteristics of Different Time Granularities

作　　者：宋丽梅 SONG Li-mei(Yangling Vocational and Technical College,Yangling,Shaanxi 712100,China)

机构地区：[1]杨凌职业技术学院,陕西杨凌712100

出　　处：《杨凌职业技术学院学报》2025年第1期19-21,共3页Journal of Yangling Vocational & Technical College

基　　金：杨凌职业技术学院2024年校内科研基金项目“城市轨道交通列车跳站与客流控制协同优化研究”(ZK24-15)。

摘　　要：地铁客流在线网中的时空分布统计预测,便于高效匹配客流需求,保障城市轨道交通运输能力,为探究地铁进站客流量预测精度与其时间粒度二者之间的关系,提出了一种基于不同时间粒度的耦合时空特征下短时客流预测方法,选取西安地铁进站客流量数据进行研究,将有效运营时间划分为不同时间粒度,分别是0.5 h、1 h以及1 d,采用Pearson系数法对不同时间粒度下的客流量时间序列进行预测,利用ARIMA模型(差分整合移动平均自回归)对全网进站流量进行拟合分析,结果表明:时间粒度为0.5 h、1 h以及1 d时,ARIMA模型预测结果平均相对误差分别为4.58%、3.86%、4.12%,平均相对误差最小的是1 h,最大的是1 d,不同时间粒度下预测误差变化趋势大致相同,对短时客流预测具有较高精度的是优化后的时间序列模型,预测结果可以实现运营组织优化,实现提高轨道交通运输效率、改善运营服务质量的目的。The statistical prediction of time and space distribution in the subway passenger flow online network is convenient for efficiently matching passenger flow demand and ensuring urban rail transportation capacity.In order to explore the relationship between the prediction accuracy of subway inbound passenger flow and its time granularity,a method is put forward based on the short-term passenger flow prediction method under the coupled spatio-temporal characteristics of different time granularities,the passenger flow data of Xi'an subway station is selected for research,the effective operating time is divided into different time granularities,respectively 0.5h,1h and 1d,and the Pearson coefficient method is used to predict the different passenger flow time series under the time granularity,and the ARIMA model(differential integrated moving average autoregressive)is used to fit and analyze the inbound flow of the whole network.The results show that:when the time granularity is 0.5h,1h,and 1d,the average relative errors of ARIMA model predicting results were 4.58%,3.86%,and 4.12%,respectively.The smallest average relative error was 1h and the largest was 1d.The change trend of forecast errors at different time granularities was roughly the same.The short-term passenger flow forecasting had higher accuracy.After the optimized time series model,the forecast results can realize the optimization of operation organization,realize the purpose of improving the efficiency of rail transportation and improving the quality of operation services.

关 键 词：城市轨道交通 短时客流预测 时间粒度 耦合时空特征 ARIMA模型 

分 类 号：U293.13[交通运输工程—交通运输规划与管理]