Wind speed prediction and energy estimation using the SARIMA method in Banyumas Regency

Abstract

Electricity consumption in Banyumas Regency shows a significant upward trend, indicating growing energy needs across various sectors. Dependence on fossil fuels poses challenges, including environmental pollution, limited resources, and price fluctuations. As a strategic solution, developing new and renewable energy, especially wind energy, is crucial to achieving energy independence and environmental sustainability. This study aims to analyze and predict wind speed in Banyumas Regency and calculate the potential electricity production that residential-scale wind turbines can generate. The method used is the seasonal auto regressive integrated moving average (SARIMA). This study applies it within a machine learning framework, using a grid search for hyperparameter tuning, to accurately predict wind speed from historical NASA POWER data. The results show that the SARIMA (1, 0, 0)×(0, 1, 1, 52) model is the optimal model with the best prediction accuracy, as evidenced by the root mean squared error (RMSE) value of 0.516 m/s and the mean absolute error (MAE) of 0.441 m/s. Based on the model, the predicted average wind speed for the next three months is 3.41 m/s, potentially generating an average daily electricity output of 1.44 kWh. These results indicate that Banyumas Regency has promising potential for the development of small-scale wind power plants to support household energy needs or public street lighting.