基于ARIMA-SVM的城际高铁客流量短期预测  

Short-term Prediction of Intercity High-speed Railway Passenger Flow Using ARIMA and SVM

作　　者：耿立艳 张占福 李达 GENG Liyan;ZHANG Zhanfu;LI Da(School of Economics and Management,Shijiazhuang Tiedao University,Shijiazhuang 050043,China;Nanchang Institute of Technology,Nanchang 330044,China;Sifang College,Shijiazhuang Tiedao University,Shijiazhuang 051132,China)

机构地区：[1]石家庄铁道大学经济管理学院,石家庄050043 [2]南昌理工学院,南昌330044 [3]石家庄铁道大学四方学院,石家庄051132

出　　处：《交通与运输》2020年第6期42-45,共4页Traffic & Transportation

基　　金：国家自然科学基金青年项目(61503261);2019年度河北省人才培养工程项目(A201901048);2019年中国物流学会、中国物流与采购联合会面上研究课题(2019CSLKT3-020)。

摘　　要：为提高城际高铁客流量的短期预测精度,提出一种自回归差分移动平均(ARIMA)和支持向量机(SVM)相结合的城际高铁客流量组合预测模型(ARIMA-SVM模型)。利用ARIMA模型预测城际高铁客流量的线性特征,通过SVM修正ARIMA模型的预测残差。运用ARIMA-SVM模型、ARIMA模型和SVM分别预测某高铁站城际高铁客流量周数据,根据平均绝对误差(MAE)和平均百分比误差(MPE)2个指标比较3个模型的预测性能。结果表明,ARIMA-SVM模型的MAE和MPE值明显小于ARIMA模型和SVM的对应值。此外,相比于ARIMA模型和SVM,ARIMA-SVM模型更准确地预测城际高铁客流量的非线性变动特征。因此,ARIMA-SVM模型有效提高城际高铁周客流量的短期预测精度。Based on autoregressive integrated moving average(ARIMA)model and support vector machine(SVM),this paper presented a combined prediction model,ARIMA-SVM,for improving the short-term prediction accuracy of the passenger flow of intercity high-speed rail.ARIMA model was applied to predicting the linear characteristics of the passenger flow of intercity high-speed rail.After that,the prediction residual of ARIMA model was corrected by using SVM.The weekly data of the passenger flow of intercity high-speed from a high-speed rail station was forecasted by ARIMA-SVM model,ARIMA model and SVM,respectively.The prediction performance of the three models was compared according to two indexes:mean absolute error(MAE)and average percentage error(MPE).The values of MAE and MPE of ARIMA-SVM model were significantly smaller than those of ARIMA model and SVM.In addition,compared with ARIMA model and SVM,ARIMASVM model could predict the nonlinear variation characteristics of the passenger flow of intercity high-speed rail more accurately.Therefore,ARIMA-SVM model effectively improved the short-term prediction accuracy of the weekly passenger flow of intercity high-speed rail.

关 键 词：城际高铁客流量 预测 自回归差分移动平均模型 支持向量机 

分 类 号：U491[交通运输工程—交通运输规划与管理]