基于图卷积神经网络的城市交通态势预测算法
|
闫旭,范晓亮,郑传潘,臧彧,王程,程明,陈龙彪
|
Urban traffic flow prediction algorithm based on graph convolutional neural networks
|
Xu YAN,Xiao-liang FAN,Chuan-pan ZHENG,Yu ZANG,Cheng WANG,Ming CHENG,Long-biao CHEN
|
|
表 2 基于PEMSD7(S)和PEMSD7(L)的工作日数据采用不同方法训练模型的交通流预测准确度结果 |
Tab.2 Traffic prediction performance comparison of different approaches for model training based on weekdays data of PeMSD7(S) and PeMSD7(L) |
|
算法 | MAE(15/30/45 min) | | MAPE(15/30/45 min) | | RMSE(15/30/45 min) | PeMSD7(S) | PeMSD7(L) | PeMSD7(S) | PeMSD7(L) | PeMSD7(S) | PeMSD7(L) | ARIMA | 3.635/4.069/4.462 | 3.398/3.793/4.147 | | 9.486/10.438/11.302 | 8.703/9.553/10.338 | | 8.594/9.158/9.704 | 8.133/8.632/9.131 | SVR | 4.026/4.628/5.090 | 3.830/4.433/4.864 | 12.373/13.992/15.193 | 11.873/13.272/14.264 | 8.605/9.388/10.007 | 8.344/9.142/9.709 | CNN | 3.256/3.721/3.876 | 3.292/3.417/3.436 | 7.995/9.350/10.084 | 8.182/8.652/8.814 | 5.618/6.524/6.858 | 5.928/6.254/6.294 | LSTM | 3.091/3.240/3.383 | 3.202/3.238/3.289 | 7.510/7.925/8.315 | 8.037/8.132/8.258 | 5.742/6.124/6.473 | 6.088/6.171/6.283 | STGCN | 1.878/2.564/3.052 | 1.742/2.434/2.953 | 4.359/6.233/7.560 | 4.095/5.842/7.043 | 3.839/5.440/6.454 | 3.669/5.293/6.410 | FFR-STGCN | 1.842/2.574/3.094 | 1.745/2.387/2.850 | 4.306/6.279/7.731 | 4.098/5.828/6.999 | 3.780/5.391/6.460 | 3.631/5.130/6.092 |
|
|
|