In multi-energy complementary highway energy supply systems, the understanding of electrical load variation is considered essential for scientific energy planning and accurate dynamic allocation. To improve the accuracy of long-term electrical load forecasting, a prediction model based on long short-term memory networks (LSTM) and a multi-head attention (MHA) mechanism was proposed. Firstly, feature engineering was applied to expand the dimensionality of input features. Then, an LSTM network was built to model temporal dependencies in electrical load data. The MHA mechanism was added to highlight key time-step information in the sequence. These components were combined to form the MHA-LSTM forecasting model. Finally, real electrical load data from a specific region were used to validate the model. The proposed model achieves a mean absolute percentage error (MAPE) of 3.23%, an coefficient of determination (R2) of 0.882 3, and an relative percent deviation (RPD) of 2.914 2. Improved performance is shown compared to traditional LSTM and attention-based LSTM models. Strong fitting ability and high predictive accuracy are demonstrated for long-term forecasting with complex temporal patterns.
MaiJunfeng. Research on dynamic demand forecasting for energy consumption in highway operation and maintenance[D]. Guangzhou: Guangzhou University, 2024.
[3]
CorentinK, YacineR, MonjurM. Electrical load forecasting mo-dels: a critical systematic review[J]. Sustainable Cities and Society, 2017, 35: 257-270.
JingShibo, ChenWeiwei, CaoQian, et al. Electric load forecasting based on model sample re-weighting method[J]. Chinese Journal of Electron Devices, 2024, 47(6): 1662-1668.
FangLei, ChuChengbo, HeYinghong, et al. Power load forecasting in station areas based on ASGMD-MLR-CLSTM[J]. Modern Electric Power, 2025, 42(4): 840-846.
SongYulu, FanYanfang, LiuYujia, et al. Power system state estimation method based on deep neural network-proline network hybrid model[J]. Science Technology and Engineering, 2022, 22(25): 11041-11048.
[10]
KhwajaA S, AnpalaganA, NaeemM, et al. Joint bagged-boosted artificial neural networks: Using ensemble machine learning to improve short-term electricity load forecasting[J]. Electric Power Systems Research, 2020, 179: 106080.
DuanZhongwei, ZhuMingyang, LiuFang, et al. Short-term power load forecasting based on grey wolf algorithm optimized support vector machine[J]. Journal of Wuhan Textile University, 2025, 38 (6): 59-64.
[13]
BouktifS, FiazA, OuniA, et al. Multi-sequence LSTM-RNN deep learning and metaheuristics for electric load forecasting[J]. Energies, 2020, 13: 391.
ZhouLei, LanXiaojing. Univariate short-term electrical load based on MA-CNN-LSTM-self attention[J]. Science Technology and Engineering, 2024, 24(22): 9408-9416.
[18]
ReidT C, AlfonsoF, LavazzaL, et al. A conceptual basis for feature engineering[J]. Journal of Systems and Software, 1999, 49: 3-15.
[19]
KwonB S, ParkR J, SongK B. Short-term load forecasting based on deep neural networks using LSTM layer[J]. Journalof Electrical Engineering & Technology, 2020, 15: 1501-1509.
WangYiyang. Study on flood forecasting model from the perspective of LSTM and multi-head attention mechanism[D]. Zhengzhou: North China University of Water Resources and Electric Power, 2023.
LiWei, LiXiaozhou, FanPeilin, et al. Regional electricity sales forecasting based on multi-head attention mechanism and long short-term memory network[J]. Power Demand Side Management, 2025, 27(1): 67-73.
WangQiang, LiQingyang, XiLei, et al. LSTM winter wheat water demand prediction based on multihead attention mechanism[J]. Journal of China Agricultural University, 2025, 30(4): 38-50.
TuXinning, YuXingchen, LiuDequan, et al. A study of CNN-LSTM model for coal gas concentration prediction model based on wavelet analysis[J]. Journal of North China Institute of Science and Technology, 2025, 22(3): 21-30.