Space-time traffic flow prediction method based on cross attention mechanism

The invention discloses a space-time traffic flow prediction method based on a cross attention mechanism, and the method comprises the steps: capturing the periodic characteristics of space-time traffic data through the cross attention mechanism which improves an attention mechanism; a graph convolutional network and a time convolutional network are utilized to respectively capture spatial dependency and time correlation characteristics of space-time traffic data, and then the extracted characteristics are fused, so that space-time traffic flow prediction is effectively realized. According to the method, modeling of various characteristics of the space-time traffic data is comprehensive and complete, the traffic flow prediction task can be effectively achieved, traffic flow prediction is more accurate and reliable, the requirement for data is lower, prediction accuracy and interpretability can be improved, and the method is more suitable for application and popularization. 本发明公布了一种基于交叉注意力机制的时空交通流量预测方法，通过对注意力机