Adaptive Non-Stationary PV Power Forecasting with ADWF-LSTM Networks
DOI:
https://doi.org/10.54691/9jxgvz94Keywords:
Photovoltaic Power Prediction, Adaptive Wavelet Fusion, Non-stationary Prediction, LSTM.Abstract
Photovoltaic (PV) power data exhibits strong non-stationarity due to solar irradiance fluctuations, diurnal temperature variations, and operating condition changes. Traditional Discrete Wavelet Transform (DWT)-LSTM models are limited by fixed wavelet bases and static fusion strategies, failing to effectively capture dynamic multi-scale features. To address this, we propose an Adaptive Dynamic Wavelet Fusion-LSTM (ADWF-LSTM) model, which integrates an Adaptive Dynamic Wavelet Fusion (ADWF) module with LSTM. The ADWF module dynamically optimizes wavelet basis parameters and multi-scale fusion weights in a data-driven manner, while LSTM retains its strength in modeling long-term temporal dependencies. Extensive experiments on the real-world Yulara PV dataset across four seasons show that ADWF-LSTM outperforms baseline models (LSTM, DWT-LSTM, TCN-LSTM, Informer). Its average MAE, MSE, and RMSE reach 0.4669, 0.5451, and 0.7275, reducing errors by over 10% on average. With an MBE of 0.0347, the model achieves optimal bias control. Visualization results confirm its superior trend-tracking and anti-interference performance under varying seasonal data characteristics.
Downloads
References
[1] Wen Y ,Wu J ,Lin P , et al.The role of offshore wind and solar PV resources in global low-carbon transition.[J].Science advances,2025,11(43):eadx5580.DOI:10.1126/SCIADV.ADX5580.
[2] Zhang W ,Zhang Z ,Dai Y , et al.Research on reactive power compensation control method for improving the voltage stability of photovoltaic station area[J].Frontiers in Energy Research,2024,121396074-1396074.DOI:10.3389/FENRG.2024.1396074.
[3] Ghali Y ,Sathyajith M ,Joao L .Direct and indirect short-term aggregated turbine- and farm-level wind power forecasts integrating several NWP sources[J].Heliyon,2023,9(11):e21479-e21479.DOI:10.1016/J.HELIYON.2023.E21479.
[4] ZouH ,YangC ,MaH , et al.Short‐term power prediction of distributed PV based on multi‐scale feature fusion with TPE‐CBiGRU‐SCA[J].IET Generation, Transmission & Distribution,2024,18(20):3200-3220.DOI:10.1049/GTD2.13266.
[5] Chaoui A K ,Fadil E H ,Choukai O , et al.A Wavelet–Attention–Convolution Hybrid Deep Learning Model for Accurate Short-Term Photovoltaic Power Forecasting[J].Forecasting,2025,7(3):45-45.DOI:10.3390/FORECAST7030045.
[6] Shen A ,Lin Y ,Peng Y , et al.DSC-CBAM-BiLSTM: A Hybrid Deep Learning Framework for Robust Short-Term Photovoltaic Power Forecasting[J].Mathematics,2025,13(16):2581-2581.DOI:10.3390/MATH13162581.
[7] Liu Y ,Guan L ,Hou C , et al.Wind Power Short-Term Prediction Based on LSTM and Discrete Wavelet Transform[J].Applied Sciences,2019,9(6):1108.DOI:10.3390/app9061108.
[8] Yousaf M Z , Singh A R , Khalid S ,et al.Bayesian-optimized LSTM-DWT approach for reliable fault detection in MMC-based HVDC systems[J].Scientific Reports, 14(1):1-24[2025-12-13].DOI:10.1038/s41598-024-68985-5.
[9] Das K U ,Tey S K ,Seyedmahmoudian M , et al.Forecasting of photovoltaic power generation and model optimization: A review[J].Renewable and Sustainable Energy Reviews,2018,81912-928. https://doi.org/10.1016/j.rser.2017.08.017.
[10] Karol ,César A ,Víctor E , et al.Intra-hour photovoltaic forecasting through a time-varying Markov switching model[J].Energy,2023,278(PB). https://doi.org/10.1016/j.energy.2023.127952.
[11] Laurentiu F ,Alexandru D ,Dan C , et al.Forecasting of Energy Production for Photovoltaic Systems Based on ARIMA and ANN Advanced Models[J].International Journal of Photoenergy,2021. https://doi.org/10.1155/2021/6777488.
[12] Dahidi A S ,Alrbai M ,Alahmer H , et al.Enhancing solar photovoltaic energy production prediction using diverse machine learning models tuned with the chimp optimization algorithm[J].Scientific Reports,2024,14(1):18583-18583. https://doi.org/10.1038/s41598-024-69544-8.
[13] Lwesya F ,Kibambila V .A Comparative Analysis of the Application of Seasonal ARIMA and Exponential Smoothing methods in short run Forecasting Tourist Arrivals in Tanzania[J].European Journal of Business and Management,2017,9(10):56-69.
[14] Dadi P S , Meera R , Rajkumar K ,et al.Enhance Energy Efficiency based Automobile Wheels by using Piezoelectric Energy Harvesting Technology and Sustainable Power Generation[C]//2024 2nd International Conference on Networking and Communications (ICNWC).[2025-10-29].DOI:10.1109/ICNWC60771.2024.10537585.
[15] Pierro M ,De Felice M ,Maggioni E , et al. Data-driven upscaling methods for regional photovoltaic power estimation and forecast using satellite and numerical weather prediction data[J].Solar Energy,2017,158:1026–1038. https://doi.org/10.1016/j.solener.2017.09.068.
[16] Lee H ,Lee BT.Confidence-aware deep learning forecasting system for daily solar irradiance[J]. IET Renew Power Gener,2019,13(10):1681–1689. https://doi.org/10.1049/iet-rpg.2018.5354.
[17] Alonso-Montesinos J ,Batlles F ,Portillo C.Solar irradiance forecasting at one-minute intervals for different sky conditions using sky camera images[J]. Energy Convers Manag,2015,105:1166–1177. https://doi.org/10.1016/j.enconman.2015.09.001.
[18] Geraldi E ,Romano F ,Ricciardelli E.An advanced model for the estimation of the surface solar irradiance under all atmospheric conditions using msg/seviri data[J]. IEEE Trans Geosci Remote Sens,2012,50(8):2934–2953. https://doi.org/10.1109/TGRS.2011.2178855.
[19] Perez R ,Lorenz E ,Pelland S , et al.Comparison of numerical weather prediction solar irradiance forecasts in the US, Canada and Europe[J]. Solar Energy,2013,94:305–326.
[20] Bamisile O , Oluwasanmi A , Obiora S ,et al.Application of deep learning for solar irradiance and solar photovoltaic multi-parameter forecast[J].Energy Sources Part A Recovery Utilization and Environmental Effects, 2020.DOI:10.1080/15567036.2020.1801903.
[21] Zhang J ,Hao Y ,Fan R , et al.An Ultra-Short-Term PV Power Forecasting Method for Changeable Weather Based on Clustering and Signal Decomposition[J].Energies,2023,16(7):DOI:10.3390/EN16073092.
[22] Lyu J ,Zhu G ,He C .Data-driven short-term photovoltaic power forecasting under extreme weather conditions using GRU-KAN model[J].Renewable Energy,2026,257124756-124756.DOI:10.1016/J.RENENE.2025.124756.
[23] Aouni E A ,Naimi E S ,Ayat Y .Machine learning-based photovoltaic power and energy prediction in time–frequency domain[J].Electrical Engineering,2025,107(11):1-12.DOI:10.1007/S00202-025-03272-8.
[24] Guo Y ,Song Y ,Lai Z , et al.Learning Coupled Meteorological Characteristics Aids Short-Term Photovoltaic Interval Prediction Methods[J].Energies,2025,18(2):308-308.DOI:10.3390/EN18020308.
[25] Ospina J ,Newaz A ,Faruque O M .Forecasting of PV plant output using hybrid wavelet-based LSTM-DNN structure model[J].IET Renewable Power Generation,2019,13(7):1087-1095.DOI:10.1049/iet-rpg.2018.5779.
[26] Deng R ,Wang Y ,Xu P , et al.A high-precision photovoltaic power forecasting model leveraging low-fidelity data through decoupled informer with multi-moment guidance[J].Renewable Energy,2025,250123391-123391.DOI:10.1016/J.RENENE.2025.123391.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Frontiers in Science and Engineering
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
