基于移动通信数据分析的Elman神经网络城市轨道交通客流预测

城市轨道交通的短时客流预测数据对运营组织单位面对潜在的大客流或突发事件的应对准备工作有着重要的作用。以原始移动通信数据作为换乘站点换乘客流统计的数据来源,得到了精确的单条线路某个换乘站的换乘人数,并结合自动售检票系统的统计数据,通过建立Elman神经网络模型对客流数据进行样本对训练,得到下游车站未来1 h内断面客流量的预测结果。预测结果误差符合要求,为站点的运营组织方案提供了良好的数据支撑。同时为了对比说明建立了ARIMA模型,并对预测结果作出分析比较。

Elman Neural Network Prediction of Rail Transit Passenger Flow Based on of Mobile Data Analysis

Short term passenger flow prediction data plays a significant role in urban rail transit operation and emergency events handling for operators.In this article;the original mobile communication data for passenger flow statistics are used to present the passenger flow at a transfer station and attain the precise amount of the interchange station passenger flow.Through the establishment of an Elman neural network;sample pairs training is conducted together with the AFC data in order to acquire the sectional passenger flow at the next station in the following hour.The prediction error meets the requirements and provides the operation organization scheme with solid confirmation.Meanwhile;ARIMA (autoregressive integrated moving average) model is developed to make a comparison with Elman network prediction results.