基于ConvLSTM的高速公路交通流预测仿真研究

Simulation Study on Highway Traffic Flow Prediction Based on ConvLSTM

交通流通常具有复杂时空关联性,且易受天气、速度等外部因素的影响。为提高高速公路关键节点交通流预测的准确性,设计一种基于ConvLSTM网络且融合时空关联性和外部因素的交通流预测模型——STE-ConvLSTM。构建交通流、速度、天气时空矩阵,将其延深度方向堆叠,通过滑动窗口模型将其处理为类图像时间序列数据,利用ConvLSTM网络提取交通流的时空关联性和外部因素特征;利用卷积层实现交通流预测多变量多步输出。实验结果表明,相较于传统的交通流预测模型,该模型在交通流多步预测方面的预测准确度有所提升。

Traffic flow usually has complex temporal and spatial correlations,and is susceptible to external factors such as weather and traffic speed.In order to improve the accuracy of traffic flow prediction at key nodes of the expressway,a traffic flow prediction model,STE-ConvLSTM model,based on convolutional LSTM network and integrating spatio-temporal correlation and external factors is designed.Firstly,the spatio-temporal matrix of traffic flow,speed and weather is constructed,and stacked in the direction of depth,which is processed into time series data of image-like images by sliding window model,and the characteristics of spatio-temporal correlation and external factors of traffic flow are extracted by ConvLSTM network;finally,the multi-variable multi-step output of traffic flow prediction is realized by using convolutional layers.Experimental results show that compared with the traditional traffic flow prediction model,the prediction accuracy of the model in terms of multi-step prediction of traffic flow is improved.