基于时空序列的铁路客运量数据挖掘预测

在分析铁路客运量数据的时空复杂性特征的基础上,以铁路假日运输管理系统中春运期间的客运量数据为依据,采用BP神经网络的数据挖掘方法进行铁路客运量数据挖掘预测研究。通过BP神经网络的建模方法把客运量的空间属性、数据属性和时间属性有机地结合起来,将数据的建模含于网络的数值当中。网络在学习过程中系统误差始终保持持续稳定的下降趋势,没有产生局部振荡和陷入极小现象,整个学习过程中系统稳定性较好。各样本之间的期望输出和实际输出之间吻合较好,从而证明所采取的数据处理方法的有效性和网络学习参数的合理性。根据BP神经网络得到的预测模型在仿真试验中的期望输出和实际输出之间吻合较好,预测的客运量和实际客运量数值非常接近。

Study on Prediction of Railway Passenger Traffic Volume Based on Time-space Serial

In accordance with the passenger traffic volume data of railways during spring festival of railway holiday traffic management system, the BP Neural Network is adopted to predict the passenger traffic volume of railways in data mining based on analyzing the spatio-temporal complexity character of railway passenger ticket in this paper. The spatial attribute, data attribute and time attribute of the passenger traffic volume of railways are organically syncretized by the BP neural network modeling, and the data modeling is contained in the data of the network. The system average error is gradually approaching to the anticipant system average error and the stability of the system is better in the course of learning, which does not engender local concussion and trap in local minimum. The anticipant and the actual output results of all the segments are very similar which is proved that the data processing method is effective and the parameters of BP neural network are reasonable. The simulation results show that the anticipant and the actual output results are very similar, and at the same time, the anticipant and the actual output results of the passenger traffic volume are very proximity.