A master's thesis at the College of Computer Science and Information Technology at Kirkuk University discussed deep learning for solar energy forecasting.
The thesis, submitted by student Azhin Hussein Rashid, aimed to develop and compare machine learning and deep learning models for forecasting global horizontal solar irradiance (GHI) in Kirkuk, using climate data from NASA's POWER database.
The thesis included testing six machine learning models, along with the CNN-LSTM and Transformer deep learning models, employing a walk-forward approach for time verification and data leakage prevention, in addition to feature selection and parameter optimization techniques.
The results showed the superiority of the Random Forest model.
