摘要
针对现有铁路货运量预测方法存在较大突变性误差的问题,提出经济周期阶段参数的概念,将经济周期量化后作为一个输入因素提供给神经网络模型,用以学习记忆经济波动情况,建立基于经济周期的Elman神经网络预测模型,并以我国1992~2008年铁路货运量为实例对方法进行检验,与BP神经网络预测结果进行对比。实例表明,该方法有效减小突变性误差,预测精度较高,Elman神经网络在进行动态系统预测时效果更佳。
Considering the unexpected errors of existing railway freight volume forecasting methods,the concept of the economic-cycle-phases parameter was put forward and introduced into the neural network model as a quantitative influence factor of economic cycles to study and remember the economic fluctuation and the Elman neural network model based on economic cycles was established.The improved forecasting model was tested by the data of freight volumes from 1992 to 2008 and compared with the BP neural network forecasting model.The results show that the improved method reduces unexpected errors efficaciously and is of high forecasting precision and the Elman neural network is more effective in dynamic system forecasting.
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2010年第5期1-6,共6页
Journal of the China Railway Society
基金
铁道部科技研究开发计划项目(2008X020-B
2010X014
2009F021)
教育部中央高校基本科研业务费专项资金中南大学前沿研究计划(2010QZZD021)
关键词
铁路货运量预测
经济周期
经济周期阶段参数
自组织竞争神经网络
ELMAN神经网络
railway freight volume forecasting
economic cycle
economic-cycle-phases parameter
self-organizing competitive neural network
Elman neural network