基于RBF神经网络的铁路旅客周转量预测研究  

A Study on the Prediction of Railway Passenger Volumes based on RBF Neural Network

作　　者：颜保凡[1] 郭垂江[1] 

机构地区：[1]湖南铁路科技职业技术学院运输管理学院,湖南株洲412000

出　　处：《铁道运输与经济》2017年第12期65-69,共5页Railway Transport and Economy

基　　金：湖南省教育科学"十三五"规划课题(XKJ17BZY040);湖南省高校科研项目(15C0908)

摘　　要：为实现对一定时期内旅客周转客流量进行有效预测,在分析RBF神经网络原理和铁路旅客周转量数据统计的基础上,结合时间序列归一化转化分析方法,建立基于RBF神经网络的铁路旅客周转量预测模型。选取2000—2013年实际的旅客周转量数据对神经网络模型进行训练,并用2014—2015年数据对模型精度进行可靠性检验。检验结果表明,RBF神经网络模型具有可靠的预测精度,可以有效地对铁路旅客周转量进行预测,从而更好地为铁路运输组织、站场线路设计等提供可靠的依据。With an analysis of RBF neural network model and railway passenger volumes, this paper proposes a model for the prediction of railway passenger volumes based on RBF neural network. This model aims to accurately predict the railway passenger volume in a given period by using the method of normalized conversion of time series. The passenger volumes from 2000 to 2013 are used to train the proposed prediction model and the 2014-2015 data are then used to prove the reliability of the model, which turns out to be reliable and accurate in forecasting railway passenger volumes. The findings are expected to provide a reliable theoretical basis for further studies of railway passenger volumes with the purpose of improving railway transportation of station yards and routes.

关 键 词：铁路 旅客周转量 神经网络 非线性 时间序列 运量预测 

分 类 号：U293.1[交通运输工程—交通运输规划与管理]