Prediction and classification of solar photovoltaic power generation using extreme gradient boosting regression model

Solar energy is well-positioned for adoption due to the aggregate demand for renewable energy sources and the reduced price of solar panels. Solar photovoltaic (PV) electricity has many benefits over wind power, including lower noise levels, quicker installation, and more location versatility. However, there are difficulties, including the possibility of unpredictability between accessible power supply and load demand that comes with the rapid use of intermittent renewable energy sources. Hence, this study proposes the Extreme Gradient Boosting regression-based Solar Photovoltaic Power Generation Prediction (XGB-SPPGP) model to predict and classify the usage of solar power successfully with minimal error. Extreme gradient boosting regression is an effective and reliable method for solar PV power generation predictions, particularly in cases where the target-input feature relationship is complex and non-linear. Relative humidity, temperature, clear-sky index, and time of day are the most critical input features to improve the model's accuracy. A comprehensive and reliable evaluation is provided by validating the proposed model using data from various climatic locations. The model's performance is then further investigated by conducting a seasonal study. Solar energy has the potential to be a reliable and long-term part of the electrical power system's growth, and these findings have significant consequences for grid management, energy planning, and governance. With generation estimation capability, an IoT-based datalogger for a stand-alone PV panel is established. The outcomes and features acquired validate the suggested methods' superiority in forecasting electricity production. The experimental outcomes of the XGB-SPPGP model increase the power generation forecasting ratio of 99.3%, accuracy ratio of 98.7%, overall performance ratio of 97.2%, and weather prediction ratio of 95.5% and reduce mean absolute error by 8.4% compared to other popular models.