Forecasting solar power generation as a renewable energy utilizing various machine learning models

Solar power is a renewable energy that uses sunlight to generate electricity. Some solar technologies, such as photovoltaic (PV) panels, convert sunlight into electrical energy. Since solar power generation does not emit greenhouse gases or produce air pollution, it can become one of the solutions to global warming and climate change. However, solar power generation needs to be more consistent as unpredictable meteorological and environmental factors influence it. Thus, it challenges the production of consistent and efficient electrical energy through PV panels. The unpredictability of environmental and climatic conditions makes it more difficult for the government and general public to produce solar energy for everyday usage. Therefore, precise solar power generation forecasting is necessary for a renewable energy system to operate effectively and economically. In this study, various machine learning models were applied for forecasting solar power generation. The applied models were Polynomial Regression, Support Vector Regression (SVR), K-Nearest Neighbours, Random Forest, Extreme Gradient Boosting (XGBoost), and Artificial Neural Network (ANN). According to the findings, Random Forest outperformed all other models with R = 0.877 while Support Vector Regression had least performance with R² = 0.487. The predicted solar power generation is important in integrating PV panels into traditional electrical grid systems.