基于ARMA与回归修正组合方法的城市轨道交通客流预测研究  

Research on Passenger Flow Prediction of Urban Rail Transit Based on Combination Method of ARMA and Regression Correction

作　　者：廖桂妤 LIAO Guiyu(Baosight Software(Guangxi)Co.,Ltd.,Nanning 530025,China)

机构地区：[1]宝信软件(广西)有限公司,广西南宁530025

出　　处：《现代信息科技》2025年第5期139-143,共5页Modern Information Technology

摘　　要：首先,文章分析了城市轨道交通客运量的规律与趋势,并选用自回归滑动平均(ARMA)模型对客运量进行预测。其次,针对ARMA模型对未来三天以上客运量预测准确性降低的问题,采用滚动预测方法进行优化。最后,针对ARMA滚动预测对客运量变化转折点预测性能不足的问题,分析影响客运量变化的相关因素,并采用回归分析方法进行修正。结果显示,ARMA滚动预测结合回归修正的组合方法,相较于单纯的ARMA方法,不仅保留了对未来三天以上客运量预测性能的优化,还能捕捉外部因素对客运量变化的影响,提高了客流预测的准确性。Firstly,the paper analyzes the patterns and trends of urban rail transit passenger volume,and selects the Auto-Regression Moving Average(ARMA)model to predict passenger volume.Secondly,to address the issue of reduced accuracy in predicting passenger volume for the next three days or more using the ARMA model,a rolling prediction approach is adopted for optimization.Finally,aiming at the issue of insufficient predictive performance of ARMA rolling prediction for turning points in passenger volume changes,relevant factors affecting changes in passenger volume are analyzed and regression analysis methods are adopted for correction.The results show that the combination method of ARMA rolling prediction and regression correction,compared with the pure ARMA method,not only retains the optimization of passenger volume prediction performance for the next three days or more,but also captures the impact of external factors on passenger volume changes.And it improves the accuracy of passenger flow prediction.

关 键 词：客流预测 城市轨道交通 时间序列 回归分析 

分 类 号：U491.14[交通运输工程—交通运输规划与管理] U293.13[交通运输工程—道路与铁道工程] TP391.9[自动化与计算机技术—计算机应用技术]