مقایسه و مرور کاربرد مدل‌های یادگیری عمیق در پیش‌بینی تولید انرژی الکتریکی سیستم‌های فتوولتائیک با تمرکز بر LSTM و مدل‌های ترکیبی

نوع مقاله : مقاله علمی

نویسندگان
امیرعلی سیف الدین 1
مبینا کلانتری 2
محمدعلی اله ربی شیرازی 3
1 دانشیار، دانشکده مهندسی انرژی و منابع پایدار، دانشگاه تهران، تهران، ایران
2 دانشجوی کارشناسی ارشد، دانشکده مهندسی انرژی و منابع پایدار، دانشگاه تهران، تهران، ایران
3 دانشجوی دکتری، دانشکده مهندسی انرژی و منابع پایدار، دانشگاه تهران، تهران، ایران
10.22034/stme.2026.562431.1188
چکیده
پیش‌بینی دقیق تولید انرژی خورشیدی، به دلیل ماهیت متغیر تابش خورشید، شرایط جوی پویا و عدم قطعیت‌های اقلیمی، یکی از چالش‌های اصلی در مدیریت سیستم‌های انرژی تجدیدپذیر محسوب می‌شود. در این پژوهش، یک مرور سیستماتیک جامع بر 40 مطالعه‌ی منتخب منتشرشده در بازه‌ی زمانی ۲۰۱۹ تا ۲۰۲۵ انجام شد تا کارایی مدل‌های یادگیری عمیق در پیش‌بینی تولید انرژی خورشیدی مورد بررسی قرار گیرد. تمرکز اصلی بر شبکه‌های حافظه‌ی کوتاه‌ مدت–بلند مدت (LSTM) و مدل‌های ترکیبی بوده است که در سال‌های اخیر کاربرد گسترده‌ای یافته‌اند. مقایسه‌ی مدل‌ها بر اساس شاخص‌های ارزیابی خطا، شامل ریشه‌ی میانگین مربعات خطا‌‎‌‌‌ (RMSE)، میانگین قدر مطلق خطا (MAE) و میانگین درصد خطای مطلق ‍‌(MAPE)، صورت گرفت. نتایج نشان داد که مدل‌های ترکیبی در پیش‌بینی‌های کوتاه‌ مدت عملکرد بهتری داشته و خطا را تا حدود 9/2% کاهش داده‌اند، در حالی‌که مدل‌های مجهز به مکانیزم توجه در افق‌های میان‌مدت و بلند مدت دقت بیشتری ارائه کرده‌اند. علاوه بر این، استفاده از داده‌های چند منبعی، شامل اطلاعات محلی، تصاویر ماهواره‌ای و داده‌های هواشناسی، موجب بهبود دقت پیش‌بینی تا حدود 30% شده است. در نهایت، یافته‌ها بیانگر روند روبه رشد بهره‌گیری از رویکردهای چند سطحی، تلفیق داده‌های متنوع و هوشمندسازی مدل‌ها برای ارتقای قابلیت اطمینان و پایداری در مدیریت منابع انرژی خورشیدی است.
کلیدواژه‌ها
انرژی های تجدیدپذیر
پیش‌بینی انرژی خورشیدی
شبکه‌های حافظه‌ی کوتاه‌مدت–بلندمدت
‌مدل‌هایادگیری عمیق
مدل‌های ترکیبی
موضوعات

عنوان مقاله English

Comparison and Review of the Application of Deep Learning Models in Forecasting Electrical Energy Generation of Photovoltaic Systems with a Focus on LSTM and Hybrid Models

نویسندگان English

Amirali Saifoddin 1
Mobina Kalantari 2
Mohammadali Allahrabbi Shirazi 3
1 Associate Professor, School of Energy Engineering and Sustainable Resources, Head of Soft Technologies Institute, College of Interdisciplinary Science and Technology, University of Tehran, Tehran, Iran
2 M.Sc. Student, School of Energy Engineering and Sustainable Resources, College of Interdisciplinary Science and Technology, University of Tehran, Tehran, Iran
3 PhD Student, School of Energy Engineering and Sustainable Resources, College of Interdisciplinary Science and Technology, University of Tehran, Tehran, Iran
چکیده English

Accurate forecasting of solar energy generation is considered one of the main challenges in the management of renewable energy systems due to the inherent variability of solar irradiance, dynamic weather conditions, and climatic uncertainties. In this study, a comprehensive systematic review was conducted on 40 selected studies published between 2019 and 2025 to investigate the performance of deep learning models in solar power generation forecasting. The primary focus is placed on Long Short-Term Memory (LSTM) networks and hybrid models, which have gained widespread application in recent years. The comparison of models was carried out based on error evaluation metrics, including Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE).
The results indicate that hybrid models exhibit superior performance in short-term forecasting, reducing prediction errors by up to approximately 2.9%, while models equipped with attention mechanisms provide higher accuracy in medium- and long-term forecasting horizons. Furthermore, the use of multi-source data, including local measurements, satellite imagery, and meteorological data, improves forecasting accuracy by up to approximately 30%. Overall, the findings reveal a growing trend toward the adoption of multi-level approaches, diverse data integration, and intelligent modeling techniques to enhance reliability and sustainability in solar energy resource management.

کلیدواژه‌ها English

Renewable Energy
Solar Energy Forecating
Long Short-Term Memory Networks
Deep Learning Models
Hybrid Models
[1] N. L. M. Jailani et al., “Investigating the power of LSTM-based models in solar energy forecasting,” Processes, Vol. 11, No. 5, Art. no. 1382, 2023, DOI: 10.3390/pr11051382.
[2] M. S. Ibrahim, S. M. Gharghory, and H. A. Kamal, “A hybrid model of CNN and LSTM autoencoder-based short-term PV power generation forecasting,” Electrical Engineering, Vol. 106, No. 4, pp. 4239–4255, 2024, DOI: 10.1007/s00202-023-02220-8.
[3] N. Zhou, B. Shang, M. Xu, L. Peng, and G. Feng, “Enhancing photovoltaic power prediction using a CNN-LSTM-attention hybrid model with Bayesian hyperparameter optimization,” Global Energy Interconnection, Vol. 7, No. 5, pp. 667–681, 2024, DOI: 10.1016/j.gloei.2024.10.005.
[4] J. Liu, H. Zang, L. Cheng, T. Ding, Z. Wei, and G. Sun, “A transformer-based multimodal-learning framework using sky images for ultra-short-term solar irradiance forecasting,” Applied Energy, Vol. 342, Art. no. 121160, 2023, DOI: 10.1016/j.apenergy.2023.121160.
[5] Y. Yu, J. Cao, and J. Zhu, “An LSTM short-term solar irradiance forecasting under complicated weather conditions,” IEEE Access, Vol. 7, pp. 145651–145666, 2019, DOI: 10.1109/ACCESS.2019.2946057.
[6] Q. Qiu et al., “DTCformer: A temporal convolution-enhanced Autoformer with DILATE loss for photovoltaic power forecasting,” Energies, Vol. 18, No. 10, Art. no. 2450, 2025, DOI: 10.3390/en18102450.
[7] S. Xu, R. Zhang, H. Ma, C. Ekanayake, and Y. Cui, “On vision transformer for ultra-short-term forecasting of photovoltaic generation using sky images,” Solar Energy, Vol. 267, Art. no. 112203, 2024, DOI: 10.1016/j.solener.2023.112203.
[8] E. Kim and Y. S. Hau, “Advancing photovoltaic solar power forecasting by the hybrid model of CNN and DNN and the XAI with SHAP,” International Journal of Green Energy, pp. 1–12, 2025, DOI: 10.1080/15435075.2025.2553864.
[9] A. Agga, A. Abbou, M. Labbadi, Y. El Houm, and I. H. O. Ali, “CNN-LSTM: An efficient hybrid deep learning architecture for predicting short-term photovoltaic power production,” Electric Power Systems Research, Vol. 208, Art. no. 107908, 2022, DOI: 10.1016/j.epsr.2022.107908.
[10] G. Wu, Y. Wang, Q. Zhou, and Z. Zhang, “Enhanced photovoltaic power forecasting: An iTransformer and LSTM-based model integrating temporal and covariate interactions,” in 2024 IEEE 8th Conference on Energy Internet and Energy System Integration (EI2), 2024, pp. 856–861, DOI: 10.1109/EI264398.2024.10990816.
[11] E. M. Al-Ali et al., “Solar energy production forecasting based on a hybrid CNN-LSTM-transformer model,” Mathematics, Vol. 11, No. 3, Art. no. 676, 2023, DOI: 10.3390/math11030676.
[12] D. Salman, C. Direkoglu, M. Kusaf, and M. Fahrioglu, “Hybrid deep learning models for time series forecasting of solar power,” Neural Computing and Applications, Vol. 36, No. 16, pp. 9095–9112, 2024, DOI: 10.1007/s00521-024-09558-5.
[13] V. Esen, B. Coban, B. Y. Kavus, T. K. Karaca, T. Dindar, and A. S. Sarkin, “An interpretable statistical approach to photovoltaic power forecasting using factor analysis and ridge regression,” Scientific Reports, Vol. 15, No. 1, Art. no. 38947, 2025, DOI: 10.1038/s41598-025-22838-x.
[14] M. C. Krstevska, P. Krstevski, M. Srbinovska, V. Andova, and A. K. Mateska, “Smart forecasting: Enhancing virtual power plant performance with analytical frameworks,” e-Prime—Advances in Electrical Engineering, Electronics and Energy, Vol. 12, Art. no. 101003, 2025, DOI: 10.1016/j.prime.2025.101003.
[15] M. Husein, E. J. Gago, B. Hasan, and M. C. Pegalajar, “Towards energy efficiency: A comprehensive review of deep learning-based photovoltaic power forecasting strategies,” Heliyon, Vol. 10, No. 13, Art. no. e33419, 2024, DOI: 10.1016/j.heliyon.2024.e33419.
[16] S.-C. Lim, J.-H. Huh, S.-H. Hong, C.-Y. Park, and J.-C. Kim, “Solar power forecasting using CNN-LSTM hybrid model,” Energies, Vol. 15, No. 21, Art. no. 8233, 2022, DOI: 10.3390/en15218233.
[17] J. Yu et al., “Deep learning models for PV power forecasting: Review,” Energies, Vol. 17, No. 16, Art. no. 3973, 2024, DOI: 10.3390/en17163973.
[18] L. Cao, H. Yang, C. Zhou, S. Wang, Y. Shen, and B. Yuan, “Photovoltaic short-term output power forecast model based on improved complete ensemble empirical mode decomposition with adaptive noise–kernel principal component analysis–long short-term memory,” Energies, Vol. 17, No. 24, Art. no. 6365, 2024, DOI: 10.3390/en17246365.
[19] B. Ladjal et al., “Hybrid deep learning CNN-LSTM model for forecasting direct normal irradiance: A study on solar potential in Ghardaia, Algeria,” Scientific Reports, Vol. 15, No. 1, Art. no. 15404, 2025, DOI: 10.1038/s41598-025-94239-z.
[20] J. Wu, Y. Zhao, R. Zhang, X. Li, and Y. Wu, “Application of three Transformer neural networks for short-term photovoltaic power prediction: A case study,” Solar Compass, Vol. 12, Art. no. 100089, 2024, DOI: 10.1016/j.solcom.2024.100089.
[21] G. Wu, D. Hu, Y. Zhang, G. Bao, and T. He, “A convolutional neural network–long short-term memory–attention solar photovoltaic power prediction–correction model based on the division of twenty-four solar terms,” Energies, Vol. 17, No. 22, Art. no. 5549, 2024, DOI: 10.3390/en17225549.
[22] C. Pan, Y. Liu, Y. Oh, and C. Lim, “Short-term photovoltaic power forecasting using PV data and sky images in an auto cross modal correlation attention multimodal framework,” Energies, Vol. 17, No. 24, Art. no. 6378, 2024, DOI: 10.3390/en17246378.
[23] C. Corsi, M. J. Blanco, V. Grigoriev, and J. Pye, “Upper limits to the mean annual optical efficiency of solar mono-tower systems,” Solar Energy, Vol. 236, pp. 88–99, 2022, DOI: 10.1016/j.solener.2022.02.038.
[24] R. Ding et al., “Application of machine learning in optimizing proton exchange membrane fuel cells: A review,” Energy and AI, Vol. 9, Art. no. 100170, 2022, DOI: 10.1016/j.egyai.2022.100170.
[25] T. Zhu, Y. Guo, Z. Li, and C. Wang, “Solar radiation prediction based on convolution neural network and long short-term memory,” Energies, Vol. 14, No. 24, Art. no. 8498, 2021, DOI: 10.3390/en14248498.
[26] G. Piantadosi, S. Dutto, A. Galli, S. De Vito, C. Sansone, and G. Di Francia, “Photovoltaic power forecasting: A Transformer based framework,” Energy and AI, Vol. 18, Art. no. 100444, 2024, DOI: 10.1016/j.egyai.2024.100444.
[27] Y. Huang and Y. Wu, “Short-term photovoltaic power forecasting based on a novel Autoformer model,” Symmetry, Vol. 15, No. 1, Art. no. 238, 2023, DOI: 10.3390/sym15010238.
[28] M. Awais et al., “Short-term photovoltaic energy generation for solar powered high efficiency irrigation systems using LSTM with spatio-temporal attention mechanism,” Scientific Reports, Vol. 14, No. 1, Art. no. 10042, 2024, DOI: 10.1038/s41598-024-60672-9.
[29] İ. F. Şener and İ. Tuğal, “Optimized CNN-LSTM with hybrid metaheuristic approaches for solar radiation forecasting,” Case Studies in Thermal Engineering, Vol. 72, Art. no. 106356, 2025, DOI: 10.1016/j.csite.2025.106356.
[30] M. A. Atiea, A. A. Abdelghaffar, H. B. Aribia, F. Noureddine, and A. M. Shaheen, “Photovoltaic power generation forecasting with Bayesian optimization and stacked ensemble learning,” Results in Engineering, Vol. 26, Art. no. 104950, 2025, DOI: 10.1016/j.rineng.2025.104950.
[31] W.-C. Tsai, C.-S. Tu, C.-M. Hong, and W.-M. Lin, “A review of state-of-the-art and short-term forecasting models for solar PV power generation,” Energies, Vol. 16, No. 14, Art. no. 5436, 2023, DOI: 10.3390/en16145436.
[32] M. Shafiuzzaman et al., “Enhanced very short-term load forecasting with multi-lag feature engineering and prophet-XGBoost-CatBoost architecture,” Energy, Vol. 335, Art. no. 137981, 2025, DOI: 10.1016/j.energy.2025.137981.
[33] C. Tian, J. Ma, C. Zhang, and P. Zhan, “A deep neural network model for short-term load forecast based on long short-term memory network and convolutional neural network,” Energies, Vol. 11, No. 12, Art. no. 3493, 2018, DOI: 10.3390/en11123493.
[34] B. Lim, S. Ö. Arık, N. Loeff, and T. Pfister, “Temporal fusion transformers for interpretable multi-horizon time series forecasting,” International Journal of Forecasting, Vol. 37, No. 4, pp. 1748–1764, 2021, DOI: 10.1016/j.ijforecast.2021.03.012.
[35] A. Jakoplić, D. Franković, J. Havelka, and H. Bulat, “Short-term photovoltaic power plant output forecasting using sky images and deep learning,” Energies, Vol. 16, No. 14, Art. no. 5428, 2023, DOI: 10.3390/en16145428.
[36] A. Zameer, F. Jaffar, F. Shahid, M. Muneeb, R. Khan, and R. Nasir, “Short-term solar energy forecasting: Integrated computational intelligence of LSTMs and GRU,” PLOS ONE, Vol. 18, No. 10, Art. no. e0285410, 2023, DOI: 10.1371/journal.pone.0285410.
[37] L. N. Boucetta, Y. Amrane, A. Chouder, S. Arezki, and S. Kichou, “Enhanced forecasting accuracy of a grid-connected photovoltaic power plant: A novel approach using hybrid variational mode decomposition and a CNN-LSTM model,” Energies, Vol. 17, No. 7, Art. no. 1781, 2024, DOI: 10.3390/en17071781.
[38] Y. Mariappan, K. Ramasamy, and D. Velusamy, “An optimized deep learning based hybrid model for prediction of daily average global solar irradiance using CNN SLSTM architecture,” Scientific Reports, Vol. 15, No. 1, Art. no. 10761, 2025, DOI: 10.1038/s41598-025-95118-3.
[39] A. Naziri, A. R. Tahavvor, M. A. Shirazi, and R. Zahedi, “Feasibility study on heat recovery from gas turbine exhaust for absorption chiller operation and efficiency enhancement using neural networks,” Thermal Science and Engineering Progress, Vol. 67, Art. no. 104210, 2025, DOI: 10.1016/j.tsep.2025.104210. (in Persian)
[40] A. Saifoddin, N. Mirzaei, M. Allahrabbi Shirazi, and H. Yousefi, “Comparative applications of supervised and unsupervised machine learning models in energy systems,” Journal of Energy Management and Technology, Vol. 9, No. 4, pp. 284–290, 2025, DOI: 10.22109/jemt.2025.547118.1573. (in Persian)
‏
 
 

  • تاریخ دریافت 05 آذر 1404
  • تاریخ بازنگری 17 دی 1404
  • تاریخ پذیرش 12 اردیبهشت 1405
  • تاریخ اولین انتشار 12 اردیبهشت 1405
  • تاریخ انتشار 01 مرداد 1405