PHOTOVOLTAIC GENERATION FORECASTING CONSIDERING HEATWAVES VIA XGBOOST WITH ENHANCED TEMPERATURE AND IRRADIANCE ESTIMATION
Palavras-chave:
Impacts, Public Policies, and Energy Systems PlanningResumo
Heat waves, which are increasingly frequent due to climate change, negatively impact photovoltaic generation by reducing module efficiency. This paper presents an enhanced forecasting method based on the XGBoost (Extreme Gradient Boosting) algorithm to improve the estimation of air temperature and solar irradiance during extreme heat events. A binary indicator, derived from historical thresholds, is integrated into the model to explicitly capture the effects of heat waves on photovoltaic performance. Using meteorological data from Santo André, Brazil, results show a reduction in mean absolute error for temperature from 1.71 °C to 1.64 °C and for irradiance from 16.99 W/m² to 16.53 W/m² compared to a baseline model without the indicator. These findings highlight the potential of incorporating extreme climate indicators to enhance forecasting accuracy and support more reliable energy planning in the face of climate variability.
