基于灰色神经网络的铁路货运量组合预测  

Combination Prediction Model of Railway Freight Volume Based on Gray Model and Neural Network

作　　者：邵梦汝 程天伦 马晓晨[1] 

机构地区：[1]西南交通大学交通运输与物流学院,成都610031 [2]西南交通大学综合交通运输智能化国家地方联合工程实验室,成都610031

出　　处：《交通运输工程与信息学报》2016年第3期129-135,共7页Journal of Transportation Engineering and Information

摘　　要：组合预测对于不完备信息的预测具有一定的实用性。鉴于铁路货运量的复杂性和非线性特征,利用我国铁路货运量的历史数据,采用无偏GM(1,1)、WPGM(1,1)、PGM(1,1)模型,将灰色预测模型与BP神经网络模型组合,构建了灰色-神经网络的组合模型。这一模型的构建对于复杂系统中不完备信息的处理有较高的实用价值。实例应用该方法对往年铁路货运量进行预测,结果表明基于灰色-神经网络组合模型在模拟性能和预测准确度方面有很大程度的提高,可以作为铁路货运量预测的有效工具,同时,为今后铁路运输生产中货运量的预测提供理论依据。Combination forecast method has practicality function in some degrees for incompleteinformation prediction. According to the nonlinear characteristics and complexity of railway freight volume, using the history data of our country railway freight volume and combining the grey prediction model with BP neural network model, a grey-neural network combination model was constructed by adoptingGM（1,1）,WPGM（1,1）, andPGM（1,1） models. Some empirical analysis show that this model can get much accurate prediction effects, it can be used to predict railway volume in the future. At the same time, the railway freight volume forecast results can be used to improve the railway organization work, and to provide the theory support for increasing railway transport capacity.

关 键 词：灰色神经网络 铁路货运量 组合预测 

分 类 号：U294.1[交通运输工程—交通运输规划与管理]