Publication

Abstract : The growing global demand for renewable energy sources has positioned Solar energy as a vital and key factor in the move towards sustainable energy systems. In particular, Solar Photo Voltaic (PV) panels have been demonstrated as possible sustainable energy systems but are affected by environmental factors like irradiance, humidity and temperature that can create uncertainty in their means of producing energy. To address this, Machine Learning (ML) algorithms are employed to forecast the solar energy produced and to help understand this uncertainty and improve solar energy modelling. In this study five separate multiple regression based ML models, Decision Tree (DT), Random Forest (RF), Gradient Boosting (GB), Light Gradient Boosting Machine (Light GBM) and Extreme Gradient Boosting (XGBoost) are used to develop predictions of solar energy production based on environmental input data. Results confirmed that ensemble models, particularly XGBoost and RF, outperformed others in predictive generalization and accuracy in terms of lowest value of Root Mean Squared Error (RMSE), Mean Absolute Error (MAE) and the highest value of R-squared(R2), which are accepted as one of the most important performance criteria by the ML models.