Wind Speed Forecasting Using Machine Learning Models: A Portuguese Wine Farm Case Study

With the European Union's requirement for reducing the amount of energy generated from non-renewable sources, there is a need for increased production of energy from renewable sources such as solar and wind power, among others. Due to the stochastic nature of natural resources that serve as these renewable energy sources, it necessitates adaptation by electrical energy systems. Predicting these resources is crucial for better planning and management of electrical energy systems. This paper aims to forecast wind speed using machine learning models, specifically comparing AutoRegressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) models. The results show that the LSTM is able to reach a Root Mean Square Error (RMSE) and a Mean Absolute Error <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(M A E)$</tex> of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{3. 1 4 5}$</tex> and 2.245, respectively, while the ARIMA achieves a higher error of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{3. 4 6 0}$</tex> and 3.031, respectively. The results allows to conclude that the LSTM model shows a more effective performance, with a lower error rate, due to its ability to recognize patterns over longer periods.