Construction of the Prediction Model of Traffic Flow by Using Computer Deep Learning Algorithm and Its Application in Logistics Management


  • Yanqi Zhang, Yuqing Fan, Zhigang Song, Zhichao Wang, Yunjiao Xu


Through the deep learning algorithm of computers, the prediction model of traffic flow in the road network is constructed to predict the traffic flow in the network more accurately and efficiently, and solve the “last kilometer” problem in logistics distribution. Firstly, the prediction model of traffic flow is studied by using graph convolution network (GCN) in deep learning algorithm. In the process of prediction, the improved track graph convolutional network (TGCN) model is introduced to better process data in deep layers and predict the complex traffic flow. The efficiency of data processing and prediction accuracy of the TGCN model are analysed by using the traffic flow data and prediction parameters in the actual road network. The mean absolute percentage error (MAPE) of the TGCN model is 4.8%, indicating that the accuracy of the TGCN model is high. Compared with other models of deep learning methods, the TGCN model is proved to have great advantages in the accuracy of data prediction. In the actual logistics transportation and distribution, the prediction model of traffic flow by using deep learning algorithm has better prediction performance.