时间序列和神经网络下我国铁路客运量的预测研究  

On Forecasting China Railway Passenger Volume Based on Time Series and Neural Network

作　　者：王国贤 范英兵 王凤玲 谢安琪 Wang Guoxian;Fan Yingbing;Wang Fengling;Xie Anqi(School of Educational Science,Heihe University,Heihe 164300,China;School of Science,Heihe University,Heihe 164300,China)

机构地区：[1]黑河学院教育科学学院,黑龙江黑河164300 [2]黑河学院理学院,黑龙江黑河164300

出　　处：《黑河学院学报》2021年第5期182-185,共4页Journal of Heihe University

基　　金：黑河学院科研项目“时间序列分析在股指建模及预测中的应用”(KJZ202101)。

摘　　要：提高我国铁路客运量的预测精度有利于国家对交通进行规划和管理。选取2010年1月至2019年12月铁路客运量的月度数据,建立SARIMA和RBF神经网络集合组成的预测模型,此模型优点是比单一的季节性ARIMA模型和RBF神经网络模型预测精度相对提高了2.66%和1.82%,组合预测模型精确度达到0.4%以下。预测2020年1月的铁路客运量为31024.13万人。Improving the prediction accuracy of railway passenger volume is conducive to transportation planning and management in our country.This paper selects the monthly data of railway passenger volume from January 2010 to December 2019 to establish a prediction model composed of SARIMA and RBF neural network.The advantage of this model is that the prediction accuracy of the single seasonal ARIMA model and RBF neural network model gets increased by 2.66%and 1.82%,and the accuracy of the combined forecasting model reaches below 0.4%.It is predicted that the railway passenger volume in January 2020 will be 310241300 ten thousand.

关 键 词：铁路客运量 时间序列模型 神经网络 

分 类 号：F532.3[经济管理—产业经济]