高速公路收费站通行流量预测方法

从自由流开放制式收费系统的角度分析近几年我国高速公路通行流量预测问题的新变化。根据收费站运营业务需求,提出时间趋势、空间分布和特征属性等3类通行流量预测尺度。比较时间序列的经典线性预测方法和机器学习预测方法,提出采用长短期记忆(Long Short-Term Memory,LSTM)网络构建通行流量时间序列预测模型。以广东省某收费站的通行流量数据为例,利用TensorFlow中的LSTM模块给出通行流量预测的拟合结果,验证该方法的可行性。

Prediction of Traffic Flow at Highway Toll Station

The traffic flow prediction for the highway with a free flow toll system is studied.The prediction of the time trend,the spatial distribution and the attributes of vehicles is practiced.A Comparison is conducted between the linear prediction method and the machine learning prediction method,and a prediction model based on LSTM(Long Short-Term Memory)network is built.The model is verified against the LSTM module of the TensorFlow through processing the data from a toll station in Guangdong province.