Short-term forecasting of solar irradiance and temperature using deep learning models with multiple inputs and multiple outputs Çok girdili-çok değişkenli çıktıya sahip derin öğrenme modelleri ile kısa vadeli güneş ışınım şiddeti ve sıcaklık tahmini

Creative Commons License

Kaysal K., Hocaoglu F. O., ÖZTÜRK N.

Journal of the Faculty of Engineering and Architecture of Gazi University, cilt.41, sa.1, ss.463-478, 2026 (SCI-Expanded, Scopus, TRDizin) identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 41 Sayı: 1
  • Basım Tarihi: 2026
  • Doi Numarası: 10.17341/gazimmfd.1533969
  • Dergi Adı: Journal of the Faculty of Engineering and Architecture of Gazi University
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, TR DİZİN (ULAKBİM)
  • Sayfa Sayıları: ss.463-478
  • Anahtar Kelimeler: bidirectional long-short term memory, deep learning, Multi-Input/Multivariate Output Models, Solar radiation estimation, temperature estimation
  • Gazi Üniversitesi Adresli: Evet

Özet

The intermittent and fluctuating nature of solar radiation poses significant limitations for many applications. Accurate estimation of solar radiation is a crucial factor in predicting the output power of a photovoltaic system. In this study, the effects of multivariate inputs on bivariate outputs for short-term forecasts were examined, and the impact of meteorological changes on a solar power plant planned for a specific region was investigated. Additionally, the performance of various deep learning models and their hybrid combinations for predicting solar radiation and temperature was compared. Compared to other models, the M/CNN-BILSTM_II model demonstrated the best performance in estimating both temperature and radiation intensity using the three input parameters: temperature, solar radiation, and humidity. The performance of the models was evaluated using RMSE, MAE, NRMSE, SMAPE, and R² metrics. For solar radiation, these metrics were calculated as 71,21 W/m² (RMSE), 35,42 W/m² (MAE), %6,52 (NRMSE), %14,87 (SMAPE) and %94,78 (R²), respectively. For temperature values, RMSE was obtained as RMSE 0,76°C, MAE 0,54°C, NRMSE %1,59, SMAPE %12,14 and R² %99,23.