基于时空融合图卷积的交通流数据修复方法
|
侯越,韩成艳,郑鑫,邓志远
|
Traffic flow data repair method based on spatial-temporal fusion graph convolution
|
Yue HOU,Cheng-yan HAN,Xin ZHENG,Zhi-yuan DENG
|
|
表 6 不同方法的修复误差对比 |
Tab.6 Comparison of repair errors of different methods |
|
检测器 | GCN | | LGCN | | 1D-CNN_GCN | | LSTM_GCN | | STF_GCN | MAE | RMSE | | MAE | RMSE | | MAE | RMSE | | MAE | RMSE | | MAE | RMSE | detect_28 | 19.42 | 21.97 | | 18.26 | 20.93 | | 17.46 | 20.56 | | 19.54 | 22.13 | | 12.38 | 14.77 | detect_50 | 11.54 | 13.59 | | 11.31 | 13.27 | | 13.16 | 15.11 | | 12.29 | 14.47 | | 8.61 | 10.74 | detect_55 | 17.50 | 20.58 | | 17.12 | 20.05 | | 19.76 | 22.7 | | 18.45 | 21.72 | | 12.89 | 16.08 | detect_62 | 5.80 | 6.84 | | 5.69 | 6.66 | | 6.65 | 7.63 | | 6.17 | 7.26 | | 4.30 | 5.36 | $\vdots $ | $\vdots $ | $\vdots $ | | $\vdots $ | $\vdots $ | | $\vdots $ | $\vdots $ | | $\vdots $ | $\vdots $ | | $\vdots $ | $\vdots $ | PeMS08 | 26.77 | 31.69 | | 28.38 | 33.40 | | 28.72 | 33.85 | | 25.65 | 30.52 | | 24.98 | 29.95 |
|
|
|