基于重组二次分解及LSTNet-Atten的短期负荷预测

刘洪伟,王磊,刘阳,张鹏超,乔石

Short term load forecasting based on recombination quadratic decomposition and LSTNet-Atten

Hongwei LIU,Lei WANG,Yang LIU,Pengchao ZHANG,Shi QIAO

表 9 不同季节的性能比较

Tab.9 Performance comparison between different seasons

季节 tp LSTM GRU CNN-LSTM 本文模型 MAPE/% RMSE/MW MAPE/% RMSE/MW MAPE/% RMSE/MW MAPE/% RMSE/MW 春季 3 1.8732 647.8519 1.5256 504.5328 1.7768 618.2275 1.1130 375.1620 春季 6 2.6604 842.1877 2.3041 729.0185 2.1072 696.2172 1.5066 466.8072 春季 12 3.1097 978.8146 2.7551 862.4831 2.4585 783.8269 2.3577 761.4055 夏季 3 1.7202 640.5428 1.7152 584.5869 1.2806 467.5284 1.1282 401.3416 夏季 6 2.0752 776.5217 1.8062 648.3480 1.9220 699.1092 1.5260 540.9594 夏季 12 2.9253 1076.0507 2.2146 849.3715 2.4865 861.7421 1.9140 1086.1290 秋季 3 2.7708 1017.8679 2.4773 948.0372 2.4117 977.8108 1.5094 559.3695 秋季 6 3.1440 1266.9124 2.8387 1091.7896 2.8652 1097.9574 1.8423 710.4821 秋季 12 3.8081 1381.7440 4.5826 1611.7871 3.7733 1399.8027 2.9159 1061.1052 冬季 3 2.8918 892.6768 2.4245 837.6585 2.6789 859.9835 1.2748 414.5248 冬季 6 3.0045 1000.1413 2.8462 927.8010 2.8310 913.7417 1.6032 537.4815 冬季 12 3.1686 1016.1463 3.9479 1241.7799 3.9768 1217.9063 2.3122 738.8276