<![CDATA[The global transition toward renewable energy requires accurate forecasting systems to support effective planning and operational management of power generation. This study aims to analyze and forecast solar and wind energy potential using a Long Short-Term Memory (LSTM) deep learning model. The dataset consists of secondary meteorological data from July–August 2025 with an initial 5-minute resolution, resampled into hourly data. The analyzed variables include global horizontal irradiance (GHI), air temperature, and wind speed at 10 meters. Separate models were developed for solar and wind energy forecasting. Solar modeling was conducted during daylight conditions (GHI > 50 W/m²), while wind modeling utilized full 24-hour data. The solar model achieved a Mean Absolute Error (MAE) of 28.02 Watts, RMSE of 34.09 Watts, and an R² value of 0.742. Meanwhile, the wind model obtained an MAE of 4.54 W/m², RMSE of 7.07 W/m², and an R² value of 0.649. These results indicate that the LSTM approach provides good predictive performance for solar energy and moderate performance for wind energy in short-term forecasting.]]>
Copyrights © 2026
