融合知识图谱的时空多图卷积交通流量预测

李劲业,李永强

Spatial-temporal multi-graph convolution for traffic flow prediction by integrating knowledge graphs

Jinye LI,Yongqiang LI

表 1 不同交通流量预测模型对未来3个时段的交通状况预测结果

Tab.1 Prediction results of traffic conditions in next three time periods by different traffic flow prediction models

模型 t=10 min t=30 min t=60 min RMSE MAE Acc R2 RMSE MAE Acc R2 RMSE MAE Acc R2 HA 26.403 13.513 0.669 0.843 27.741 14.166 0.652 0.826 29.863 15.174 0.627 0.800 LSTM 11.874 6.709 0.833 0.966 13.243 7.433 0.815 0.954 15.218 8.454 0.796 0.935 ST-GCN 11.214 6.445 0.838 0.972 12.338 7.129 0.833 0.964 14.062 7.998 0.824 0.951 TMGCN 10.937 5.822 0.841 0.969 11.897 6.284 0.836 0.965 13.221 6.932 0.829 0.956 KST-GCN 10.625 5.673 0.849 0.971 11.421 5.962 0.843 0.966 12.637 6.371 0.834 0.959 本研究 10.013 5.226 0.872 0.978 10.816 5.536 0.864 0.972 11.853 5.953 0.853 0.963