基于CNN+LSTM的短时交通流量预测方法  

Short-term traffic flow forecasting method based on CNN+LSTM

作　　者：晏臻 于重重[1] 韩璐 苏维均[1] 刘平 YAN Zhen;YU Chong-chong;HAN Lu;SU Wei-jun;LIU Ping(School of Computer and Information Engineering,Beijing Technology and Business University,Beijing 100048,China)

机构地区：[1]北京工商大学计算机与信息工程学院

出　　处：《计算机工程与设计》2019年第9期2620-2624,2659,共6页Computer Engineering and Design

基　　金：北京市教委科技创新平台基金项目(PXM2018_014213_000033)

摘　　要：针对传统的预测方法只考虑到了交通流量的时序特征,忽略了其空间特征这一问题,提出卷积神经网络(CNN)和长短期记忆网络(LSTM)相结合的短时交通流量预测模型。通过CNN挖掘相邻路口交通流量的空间关联性,通过LSTM模型挖掘交通流量的时序特征,将提取的时空特征进行特征融合,实现短期流量预测。实验结果表明,CNN+LSTM模型预测误差明显小于其它模型,验证了考虑时空特征进行交通流量预测的有效性。A short-term traffic flow forecasting model based on convolutional neural network(CNN)and long-short-term memory network(LSTM)was proposed to solve the problem that the traditional forecasting method only considers the temporal characteristics of traffic flow,but neglects its spatial characteristics.The spatial correlation of traffic flow at adjacent intersections was mined using CNN,and the temporal features of traffic flow were mined using LSTM model.The extracted temporal and spatial features were fused to achieve short-term traffic flow prediction.Experimental results show that the prediction error using CNN+LSTM model is obviously smaller than that using other models,which verifies the validity of traffic flow forecasting considering both temporal and spatial characteristics.

关 键 词：交通流量 时空特征 预测 卷积神经网络 长短期记忆网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]