基于图神经网络的交通流预测

打开文本图片集
摘 要: 为了更好地获取交通流预测问题中的时空相关性,进一步提高预测性能,提出多分辨率时空注意力网络。该模型通过注意力图卷积操作提取交通流的动态空间特征,利用循环神经网络捕获交通流中的时间特征,再将经过上述处理后的近期交通数据与日、周交通数据进行融合与预测。
关键词: 交通流预测; 时空相关性; 图卷积网络; 注意力机制; 循环神经网络
中图分类号:TP399 文献标识码:A 文章编号:1006-8228(2022)10-09-03
Traffic flow prediction based on graph neural network
Lai Junlong
(College of Mathematics and Information, South China Agricultural University, Guangzhou, Guangdong 510642, China)
Abstract: In order to better obtain the spatio-temporal correlation of traffic flow and further improve the prediction performance, a multi-resolution spatio-temporal attention network is proposed. In this model, the dynamic spatial features of traffic flow are extracted by the attention graph convolution operation, and the temporal features of traffic flow are captured by recurrent neural network. Then, the recent traffic data processed above are fused with daily and weekly traffic data, and the fused data are used to predict the future traffic flow.
Key words: traffic flow prediction; spatial-temporal; graph convolutional network; attention mechanism; recurrent neural network
0 引言
交通流预测是智能交通系统中的重要成分,准确、有效的预测能够改善和缓解城市的交通问题。(剩余3367字)