基于主成分分析与GA-BP网络的铁路客运量预测研究

Prediction of Railway Passenger Volume Based on PCA and GA-BP Network

针对铁路客运量影响因素众多、变量之间映射关系复杂的特点,使用主成分分析方法对客运量影响因素进行处理,降低相关变量维数,消除变量间的多重共线性关系,并将转换后的变量输入到基于GA-BP的神经网络模型中,完成对于铁路客运量的预测。仿真结果表明,该模型相较于BP神经网络模型,具有更好的预测精度和更简单的结构。

Considering numerous factors influencing railway passenger volume and complex mapping relation among varia-bles, the paper firstly deals with the influencing factors with PCA （principal component analysis） method and reduces the dimension of related variables to eliminate multiple co-linear relations between variables. Then, it enters the converted vari- ables into neural network model based on GA-BP and predicts the railway passenger volume. The simulation result shows that this model is more accurate and simpler than BP neural network model.