摘要
为提高铁路货运量预测精度,针对Elman神经网络的预测精度受其权值和阀值的影响,提出了一种基于DA-Elman的铁路货运量预测方法。选择国内生产总值、铁路货运量、公路货运量、公路运营路程、铁路运营路程、铁路复线比例、铁路货物周转量和铁路运输从业人员8项指标作为DA-Elman的输入,铁路货运量作为DA-Elman的输出,建立DA-Elman的铁路货运量预测模型。选择我国2000-2018年铁路货运量数据为研究对象,研究结果表明,与PSO-Elman和Elman相比,DA-Elman的铁路货运量预测精度最高,为铁路货运量预测提供了新的方法和科学决策的依据。
In order to improve the forecasting accuracy of railway freight volume,due to the influence of weight and threshold of Elman neural network,a method for predicting railway freight volume based on DA-Elman is proposed.Eight indicators including GDP,railway freight volume,road freight volume,road operating distance,railway operating distance,railway double-track ratio,railway cargo turnover and railway transport employee are selected as DA-Elman’s input,railway freight volume as DA-Elman’s output,a model for forecasting railway freight volume based on DA-Elman is established.Compared with PSO-Elman and Elman,the results show that DA-Elman for predicting railway freight volume has the highest forecast precision,which provides a new method and scientific decision-making basis for railway freight volume forecast.
作者
宋伟
张杨
SONG Wei;ZHANG Yang(Open Education Institute,Shanxi Radio&Television University,Xi’an 710068,China;International Business Machines Corporation,Xi’an 710100,China)
出处
《微型电脑应用》
2020年第12期117-119,126,共4页
Microcomputer Applications
关键词
ELMAN神经网络
蜻蜓算法
支持向量机
铁路货运量预测
Elman neural network
Dragonfly algorithm
support vector machine
railway freight volume forecast