Transformer-LSTM-Attention Deep LearningPrediction and Multi-Time-ScaleError-Compensation Optimization forHydro-Wind-PV-Storage Clean Energy Bases UsingCascade-Hydropower Reserve
DOI:
https://doi.org/10.4108/ew.15097Keywords:
TRansformer-LSTM-attention, Deep learning prediction, Multi-time-scale optimization, Hydro-wind-PV-storage clean energy base, Net-load error compensation, Cascade hydropower reserve, Ancillary-service compensationAbstract
High wind and photovoltaic (PV) penetration propagates forecast errors across day-ahead, intraday and real-time dispatch and increases short-term regulation demand. This paper proposes a Transformer-LSTM-attention deep learning prediction and multi-time-scale error-compensation optimization strategy for a hydro–wind–PV–storage clean energy base. The prediction module generates load, wind, PV and runoff forecasts at 1 h, 15 min and 1 min resolutions by combining temporal feature extraction, sequence modeling and attention weighting. The dispatch model represents residual load, wind and PV errors as a net-load deviation and converts quantile-based, or source-wise conservative, error margins into day-ahead and intraday cascade-hydropower reserve requirements. In real-time operation, the configured hydropower reserve compensates minute-level deviations, with storage and thermal units providing supplementary regulation under ramping, operating and network constraints. A case study is conducted on a Southwest China cascade-basin hydro–wind–PV–storage base with four cascade hydropower stations and a modified 30-bus equivalent local-grid system. The day-ahead MAPE values are 5.79%, 0.79% and 0.52% for wind, PV and load. Compared with fixed reserve, the proposed strategy reduces day-ahead-to-intraday hydropower fluctuation errors from 8.42% to 4.81% and from 7.48% to 6.02%; intraday-to-real-time errors decrease from 8.27% to 6.84% and from 8.44% to 8.09%. The power-supply guarantee rate remains 100.00%, while the required storage power capacity decreases by 34.00% and 32.11%. These results show improved cross-stage dispatch consistency and lower short-term storage regulation demand under the tested high-renewable operating conditions.
Downloads
References
[1] IRENA. Renewable capacity statistics 2025 [R]. Abu Dhabi: International Renewable Energy Agency, 2025.
[2] Yu A, Lu S, O’Malia K, Prasad S. China continues to lead the world in wind and solar, with twice as much capacity under construction as the rest of the world combined [R]. Global Energy Monitor, 2024.
[3] Tian C, Tan Q, Fang G, Wen X. Hydrogen production to combat power surpluses in hybrid hydro–wind–photovoltaic power systems [J]. Applied Energy, 2024, 371: 123627. DOI: https://doi.org/10.1016/j.apenergy.2024.123627.
[4] Huang K, Liu P, Ming B, Kim J S, Gong Y. Economic operation of a wind–solar–hydro complementary system considering risks of output shortage, power curtailment and spilled water [J]. Applied Energy, 2021, 290: 116805. DOI: https://doi.org/10.1016/j.apenergy.2021.116805.
[5] Li W, He M, Wan Z, Dai F, Zhang T, Jiang Y. Long-term optimal scheduling of cascade hydro–wind–PV complementary system based on deep deterministic policy gradient [J]. Applied Sciences, 2026, 16(10): 4630. DOI: https://doi.org/10.3390/app16104630.
[6] Salman D, Direkoglu C, Kusaf M, Fahrioglu M. Hybrid deep learning models for time series forecasting of solar power [J]. Neural Computing and Applications, 2024, 36: 9095–9112. DOI: https://doi.org/10.1007/s00521-024-09558-5.
[7] Yu W, Li S, Zhang H, Kang Y, Li H, Dong H. Ultra-short-term wind-power forecasting based on an optimized CNN–BiLSTM–attention model [J]. iEnergy, 2024, 3(4): 268–282. DOI: https://doi.org/10.23919/IEN.2024.0026.
[8] Bispo Junior D A, Leite G N P, Droguett E L, et al. Short-term wind power forecasting with transformer-based models enhanced by Time2Vec and efficient attention [J]. Energies, 2025, 18(23): 6162. DOI: https://doi.org/10.3390/en18236162.
[9] Gao Q. Multi-temporal scale wind power forecast-ing based on Lasso–CNN–LSTM–LightGBM [J]. EAI Endorsed Transactions on Energy Web, 2024, 11. DOI: https://doi.org/10.4108/ew.5792.
[10] Zhang Y, Zhang Y, Pan D, Sun M, Mei H, Cao W. Distributed photovoltaic power prediction considering spatiotemporal correlation and dual Attention–LSTM [J]. EAI Endorsed Transactions on Energy Web, 2025, 12. DOI: https://doi.org/10.4108/ew.7530.
[11] Wang J, Yuan W, Zhang S, Cheng S, Han L. Implementing ultra-short-term wind power forecasting without information leakage through cascade decomposition and attention mechanism [J]. Energy, 2024, 312: 133513. DOI: https://doi.org/10.1016/j.energy.2024.133513.
[12] Liu W, Mao Z. Short-term photovoltaic power forecasting with feature extraction and attention mechanisms [J]. Renewable Energy, 2024, 226: 120437. DOI: https://doi.org/10.1016/j.renene.2024.120437.
[13] Han L, Zhang R, Wang X, Dong Y. Multi-time scale rolling economic dispatch for wind/storage power system based on forecast error feature extraction [J]. Energies, 2018, 11(8): 2124. DOI: https://doi.org/10.3390/en11082124.
[14] Wang Y, Wang Y, Zhao Z, Zhou Z, Hou Z. Multi-timescale optimal operation strategy for renewable energy power systems based on inertia evaluation [J]. Energies, 2023, 16(8): 3577. DOI: https://doi.org/10.3390/en16083577.
[15] Huo J, Liu Y, Zhang Y. A multi-source multi-timescale cooperative optimiza-tion [J]. Energies, 2026, 19(3): 721. DOI: https://doi.org/10.3390/en19030721.
[16] Zhang X, Campana P E, Bi X, Egusquiza M, Xu B, Wang C, Guo H, Chen D, Egusquiza E. Capacity configuration of a hydro–wind–solar–storage bundling system with transmission constraints of the receiving-end power grid and its techno-economic evaluation [J]. Energy Conversion and Management, 2022, 270: 116177. DOI: https://doi.org/10.1016/j.enconman.2022.116177.
[17] Shi L, Duanmu C, Wu F, He S, Lee K Y. Optimal allocation of energy storage capacity for hydro–wind–solar multi-energy renewable energy system with nested multiple time scales [J]. Journal of Cleaner Production, 2024, 446: 141357. DOI: https://doi.org/10.1016/j.jclepro.2024.141357.
[18] Oureilidis K, Malamaki K-N, Gallos K, et al. Ancillary services market design in distribution networks: review and identification of barriers [J]. Energies, 2020, 13(4): 917. DOI: https://doi.org/10.3390/en13040917.
[19] Liu F, Su Y, Liu Y, Chen D, Chen W, Kuang L. Reserve allocation of high-penetration renewable energy grids considering frequency security [J]. Fron-tiers in Energy Research, 2024, 12: 1501711. DOI: https://doi.org/10.3389/fenrg.2024.1501711.
[20] Sun F, Li Q, Wang W. Frequency regulation reserve allocation for integrated hydropower plant and energy storage systems via the marginal substitution [J]. Electronics, 2025, 14(8): 1582. DOI: https://doi.org/10.3390/electronics14081582.
[21] Liu S, Wang D, Luo J, Yan J, Li Y, Li X, Li C. Research on reserve capacity optimization of hydro–wind–solar power systems based on two-stage optimization [J]. Energies, 2026, 19(10): 2324. DOI: https://doi.org/10.3390/en19102324.
[22] Hochreiter S, Schmidhuber J. Long short-term memory [J]. Neural Computation, 1997, 9(8): 1735–1780. DOI: https://doi.org/10.1162/neco.1997.9.8.1735.
[23] Tang H, Kang F, Li X, Sun Y. Short-term photovoltaic power prediction model based on feature construction and improved trans-former [J]. Energy, 2025, 320: 135213. DOI: https://doi.org/10.1016/j.energy.2025.135213.
[24] Zhang Y, Yang Q. A survey on multi-task learning [J]. IEEE Transactions on Knowledge and Data Engineering, 2022, 34(12): 5586-5609. DOI: https://doi.org/10.1109/TKDE.2021.3070203.
[25] Wang X, Han L, Wang C, Yu H, Yu X. A time-scale adaptive dispatching strategy considering the matching of time characteristics and dispatching periods of the integrated energy system [J]. Energy, 2023, 267: 126584. DOI: https://doi.org/10.1016/j.energy.2022.126584.
[26] Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need [C]. Advances in Neural Information Processing Systems (NeurIPS), 2017, 30: 5998–6008.
[27] Turk A, Wu Q, Zhang M, Oestergaard J. Day-ahead stochastic scheduling of integrated multi-energy system for flexibility synergy and uncertainty balancing [J]. Energy, 2020, 196: 117130. DOI: https://doi.org/10.1016/j.energy.2020.117130.
[28] Barbre Z, Li G. Enhanced wind energy forecasting using an extended long short-term memory model [J]. Algorithms, 2025, 18(4): 206. DOI: https://doi.org/10.3390/a18040206.
[29] Li Y, Zhai S, Yi G, Pang S, Luo X. Short-term photovoltaic power forecasting based on ICEEMDAN-TCN-BiLSTM-MHA [J]. Symmetry, 2025, 17(10): 1599. DOI: https://doi.org/10.3390/sym17101599.
[30] Cai C, Tao Y, Zhu T, Deng Z. Short-term load forecasting based on deep learning bidirectional LSTM neural network [J]. Applied Sciences, 2021, 11(17): 8129. DOI: https://doi.org/10.3390/app11178129.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Weijie Zhao, Yi Jiang, Yuxin Xie, Panfeng Guo, Lingfei Li, Zhixiong Huang
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
This is an open-access article distributed under the terms of the Creative Commons Attribution CC BY 4.0 license, which permits unlimited use, distribution, and reproduction in any medium so long as the original work is properly cited.
