Search

Traffic flow prediction is an important and challenging task in building smart city. Accurate forecasting requires consideration of spatio-temporal characteristics consisting of multiple influences such as holiday, similar node, and weather. In order to accurately capture the spatio-temporal characteristics of road network intersections, this paper proposed a prediction model based on graph convolutional neural network, temporal algorithm Prophet and Pearson correlation coefficient to achieve accurate prediction of intersection traffic considering spatial structure, similar nodes, holidays and other influencing factors. Firstly, it introduced pearson correlation coefficient to reduce the influence of similar nodes to improve the temporal algorithm for capturing temporal features. Secondly, it used the graph convolution neural network to capture spatial features; Finally, it determined the fusion weights of the graph convolutional network and the temporal algorithm by linear regression to obtain the results of spatio-temporal fusion prediction. This paper finally extracted the intersection traffic data based on the analysis of Chengdu taxi trajectory data and conducted traffic prediction experiments. The results show that the accuracy of the model proposed in this paper is better than that of most existing baseline methods, compared with the T-GCN, ASTGCN, and AGCRN models, the MAE is reduced by 1.623, 0.724, 0.161, respectively, and the accuracy is improved by 0. 144, 0. 068, and 0. 021, respectively, which verifies the effectiveness of the model in traffic intersection flow prediction. [ABSTRACT FROM AUTHOR]