摘要
随着高速铁路票价市场化改革不断推进,旅客购票选择行为也在不断地变化。准确把握旅客在预售期间对高速铁路购票的需求规律,是预售期票价浮动方案优化的基础。通过分析预售期内旅客购票过程和购票需求的形成机制,针对预售期内高速铁路购票需求集中在距发车前3 d的特点,运用Holt-Winters加法和乘法模型、SARIMA模型和LSTM模型3种典型的时间序列预测模型,以京沪高速铁路动车组列车为例,分别对同车次的平峰时期和高峰时期、不同等级座席下的二等座席及根据列车客座率划分的不同类别车次的旅客购票数进行预测。不同模型预测结果通过相同条件下3种预测模型的检验参数的比较分析,得出了SARIMA模型预测方法拟合精度更高,最适合预售期高速铁路购票需求预测的结论。
With the continuous promotion of the market-oriented reform of high speed railway fares,the behavior of passengers purchasing tickets is also constantly changing.Accurately grasping the regular pattern of passenger demand for high speed railway tickets during the pre-sale period is the basis for optimizing the dynamic price plan during the pre-sale period.This paper analyzed the process of passenger ticket purchase and the formation mechanism of ticket demand during the pre-sale period.According to the characteristics that the demand for high speed railway tickets during the presale period is concentrated in three days before the departure,Holt-Winters addition and multiplication model,SARIMA model,and LSTM model are used.By taking the high speed railway trains on the Beijing–Shanghai Line as an example,the number of passengers purchasing tickets for the same train during flat and peak periods,second-class seats under different levels of seats,and passenger load based on different types of trains was predicted.The prediction results of three different models could be obtained by comparing and analyzing the test parameters under the same conditions,and it is concluded that the SARIMA model forecasting method has higher fitting accuracy and is most suitable for predicting the demand for high speed railway tickets during the pre-sale period.
作者
刘张家璇
苗蕾
佟璐
贺振欢
LIU Zhangjiaxuan;MIAO Lei;TONG Lu;HE Zhenhuan(School of Traffic and Transportation,Beijing Jiaotong University,Beijing 100044,China)
出处
《铁道运输与经济》
北大核心
2023年第9期34-41,共8页
Railway Transport and Economy
基金
国家自然科学基金项目(72001021)
中国国家铁路集团有限公司科技研究开发计划课题(2021F017)。
关键词
铁路运输
高速铁路
预测模型
购票需求
模型比较
Rail Transport
High Speed Railway
Forecasting Models
Ticket Purchase Demand
Model Comparison