摘要
本文分析铁路客运量影响因素,利用主成分分析(PCA)消除原始铁路客运量影响因素之间的相关性,将主成分分析结果作为BP神经网络的输入,并通过增加动量项、输入数据处理、调整学习速率优化BP神经网络,提出基于PCA-BP神经网络的铁路客运量预测模型。实例研究表明,与BP神经网络相比,PCABP神经网络能有效提高铁路客运量预测精度。
In order to forecast railway passenger volume,this paper analyzes influence factors of railway passenger volume.Based on the principal component analysis(PCA) method,this paper proposes a PCA-BP neural network,which eliminates the relevant factors' correlation between original influence factors,sets the PCA results as the BP neural network's input,optimizes the BP neural network by increasing momentum item,input data processing,and adjust the learning rate.The case results indicate that,compared with the BP neural network,PCA-BP neural network can effectively improve the accuracy of railway passenger volume forecast.
出处
《综合运输》
2016年第8期43-47,73,共6页
China Transportation Review
基金
中国铁道科学研究院科研开发基金项目课题(2015YJ003)
中国铁道科学研究院科研开发基金项目课题(2015YJ140)
关键词
主成分分析
神经网络
铁路客运量
预测
principal component analysis
neural network
railway passenger volume
forecast