Spatiotemporal Joint Power Forecasting for Hydro-Wind-Solar Complementary Systems via a Hierarchical Complementary-Prior Regularized MSTD-HGFormer
DOI:
https://doi.org/10.4108/ew.15713Keywords:
hydro-wind-solar complementary system, joint power forecasting, hierarchical complementary-prior regularization, multi-scale seasonal-trend decoupling, heterogeneous graph TransformerAbstract
Accurate joint forecasting of hydro-wind-photovoltaic bases is difficult because hydropower regulation, wind randomness and photovoltaic intermittency interact through both temporal complementarity and spatial-electrical coupling. This study aims to improve source-level and bundled-power forecasting by embedding the complementary mechanism of heterogeneous renewable sources into both representation learning and model training. A hierarchical complementary-prior regularized multi-scale seasonal-trend decoupled heterogeneous graph Transformer, denoted MSTD-HGFormer-HCPR, is proposed. The model first separates the historical multi-source sequence into low-frequency trend and high-frequency fluctuation components through learnable multi-scale moving-average kernels. The two components are then encoded by parallel heterogeneous graph Transformer branches using temporal patch attention and a static-dynamic fused adjacency that combines physical topology, historical correlation and learnable latent relations. A gated fusion module produces the final forecast, while a composite loss introduces capacity-weighted system- level and correlation-aware topology-level complementary-prior constraints. Experiments on a real 17-node hydro-wind-photovoltaic base show that the proposed method reduces rRMSE by 12.9–50.6% against seven baselines, improves the correlation coefficient by 0.011–0.013 over the strongest baseline, and keeps the forecasting accuracy at or above 96.2% for hydro, wind, photovoltaic and aggregate outputs. A second, 23-node cascade-radial base with five-year records is further examined under a rolling-origin and season- stratified protocol; the aggregate rRMSE is reduced by 13.2–44.0%, and the pooled accuracies remain 96.0% for hydropower, 96.0% for wind, 96.4% for photovoltaic and 96.2% for aggregate outputs. The proposed regularized joint forecaster provides stable and accurate spatiotemporal prediction for complementary renewable bases and reduces the accumulation of source-level errors in bundled output forecasting.
Downloads
References
[1] Rahman MM, Dadon SH, He M, et al. An overview of power system flexibility: High renewable energy penetration scenarios. Energies, 2024;17(24):6393.
[2] Song J, Zhou X, Zhou Z, et al. Review of low inertia in power systems caused by high proportion of renewable energy grid integration. Energies, 2023;16(16):6042.
[3] Zhao W, Duan Y, Xu Y, et al. A Hybrid Source-Load Power Forecasting Model for Distribution Networks with High Penetration of Renewable Energy. EAI Endorsed Transactions on Energy Web, 2026;13.
[4] Ge C, Yan J, Zhang H, et al. Joint short-term power forecasting of hydro-wind-photovoltaic considering spatiotemporal delay of weather processes. Renewable Energy, 2024;237:121679.
[5] Dong L, Li Y, Xiu X, et al. An integrated ultra short term power forecasting method for regional wind-pv-hydro. Energy Reports, 2023;9: 1531-1540.
[6] Tang Y, Zhou Y, Liu P, et al. Revolutionizing solar-hydro-wind power forecasts in regional power systems with hybrid machine learning models. Solar Energy, 2025;291:113391.
[7] Peng W, Jiang Z, Yao H, et al. Short-term hydropower generation prediction model for run-of-river hydropower plants considering hydrometeorological factors. Renewable Energy, 2025;254:123790.
[8] Zhang J, Feng Z, Niu W, et al. An interpretable physics-informed artificial intelligence model for daily hydropower generation prediction in cascaded reservoirs incorporating spatiotemporal correlations. Renewable Energy, 2026;270:125952.
[9] Jonkers J, Avendano D N, Van Wallendael G, et al. A novel day-ahead regional and probabilistic wind power forecasting framework using deep CNNs and conformalized regression forests. Applied Energy, 2024;361:122900.
[10] Michael N E, Hasan S, Al-Durra A, et al. Short-term solar irradiance forecasting based on a novel Bayesian optimized deep Long Short-Term Memory neural network. Applied Energy, 2022;324:119727.
[11] Gao Q. Multi-temporal Scale Wind Power Forecasting Based on Lasso-CNN-LSTM-LightGBM. EAI Endorsed Transactions on the Energy Web, 2024;11(1).
[12] Chen H, Zhang Z, Tong S, et al. An Ultra-Short-Term Wind Power Prediction Method Based on Quadratic Decomposition and Multi-Objective Optimization. EAI Endorsed Transactions on the Energy Web, 2024;11(1).
[13] Liu S, Gong B, Cui C, et al. Photovoltaic Power Adaptive Hybrid Forecasting Model Integrated with Multi-Dimensional Error Compensation. EAI Endorsed Transactions on Energy Web, 2026;12.
[14] Wang L, He Y. M2STAN: Multi-modal multi-task spatiotemporal attention network for multi-location ultra-short-term wind power multi-step predictions. Applied Energy, 2022;324: 119672.
[15] Zhao Y, Ye L, Pinson P, et al. Correlation-constrained and sparsity-controlled vector autoregressive model for spatio- temporal wind power forecasting. IEEE Transactions on Power Systems, 2018;33(5): 5029-5040.
[16] Song Y, Tang D, Yu J, et al. Short-term forecasting based on graph convolution networks and multiresolution convolution neural networks for wind power. IEEE Transactions on Industrial Informatics, 2022; 19(2):1691-1702.
[17] Li B, Ding B, Pang W, et al. Triple-Flow Dynamic Graph Convolutional Network for Wind Power Forecasting. Symmetry, 2025;17(12):2026.
[18] Qiu H, Shi K, Wang R, et al. A novel temporal-spatial graph neural network for wind power forecasting considering blockage effects. Renewable Energy, 2024;227:120499.
[19] Shi Z, Zhang G, Lu C, et al. Dynamic Spatio-Temporal Graph- Enhanced KANformer for high-fidelity ultra-short-term wind power forecasting. Energy, 2025;337:138590.
[20] Khodayar M, Liu G, Wang J, et al. Spatiotemporal behind-the- meter load and PV power forecasting via deep graph dictionary learning. IEEE transactions on neural networks and learning systems, 2020;32(10):4713-4727.
[21] Simeunovic J, Schubnel B, Alet P J, et al. Spatio-temporal graph neural networks for multi-site PV power forecasting. IEEE Transactions on Sustainable Energy, 2021;13(2):1210-1220.
[22] Zhang Y, Zhang Y, Pan D, et al. Distributed photovoltaic power prediction considering spatiotemporal correlation and dual Attention-LSTM. EAI Endorsed Transactions on Energy Web, 2025;12.
[23] Ni C, Chen Z, Tang Y, et al. Integrated Wind-Solar-Hydro Power Prediction Method Based on Deep Neural Network. Proceedings of the 2022 4th International Conference on Electrical Engineering and Control Technologies (CEECT), Shanghai, China, 16-18 December 2022. IEEE, 2022; 210-215.
[24] Wu H, Xu J, Wang J, et al. Autoformer: Decomposition transformers with auto-correlation for long-term series fore- casting. Advances in neural information processing systems, 2021;34:22419-22430.
[25] Wang Y, Zhu J, Kang R. Destformer: A transformer based on explicit seasonal-trend decomposition for long-term series forecasting. Applied Sciences, 2023;13(18):10505.
[26] Liu Y, Hu T, Zhang H, et al. itransformer: Inverted transformers are effective for time series forecasting[C]//International conference on learning representations. 2024;2024:11116- 11140.
[27] Xu H, Zhang J, Cheng C, et al. A novel metric for evaluating hydro-wind-solar energy complementarity. Applied Energy, 2024;373:123926.
[28] Zhou H, Zhang S, Peng J, et al. Informer: Beyond effi- cient transformer for long sequence time-series forecast- ing[C]//Proceedings of the AAAI conference on artificial intel- ligence. 2021;35(12):11106-11115.
[29] Wang H, Zheng X, Lai W. Short-term power prediction of regional distributed photovoltaic power plants based on spatiotemporal information fusion. Journal of Renewable and Sustainable Energy, 2025;17(4):046103.
[30] Lu J, Xin Z, Chi W, et al. A Hybrid GCN-GRU Model with Attention Mechanism for Ultra-Short-Term Photovoltaic Power Forecasting. Proceedings of the 2025 37th Chinese Control and Decision Conference (CCDC). Xiamen, China, 16-19 May 2025; IEEE, 2025;2650-2656.
[31] Bergmeir C, Benítez JM. On the use of cross-validation for time series predictor evaluation. Information Sciences, 2012;191:192- 213. doi:10.1016/j.ins.2011.12.028.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Weijie Zhao, Yi Jiang, Yuxin Xie, Panfeng Guo, Lingfei Li, Jianxiang Huang, 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.
