Wind speed forecasting is essential for electrical energy management and planning. Wind turbine energy production varies according to fluctuations in wind speed. Unpredictable variations in wind patterns create vulnerabilities for wind power installations. To mitigate this unpredictability, an effective approach is to anticipate wind speed at specific heights, which is crucial for the operation of wind farms. Although previous work has examined different learning algorithms, researchers are still striving to find a stable model that minimizes prediction error. The objective of this study is to predict wind speed using machine learning algorithms. The results of the study highlight the remarkable predictive performance of eleven models, which are based on regression and classification methods with cross-validation. On the test data, the evaluation reveals impressive results, with coefficients of determination ranging from 0.986023 to 0.794080. The comparative study of the different algorithms shows that the proposed LightGBM (LGB) model outperforms most similar models in the literature, achieving statistically significant accuracy values. This result made it possible to evaluate the power of a wind turbine over the 12 months of the year. It shows that March generates the most energy, while August produces the least.
Mounkang et al. (Mon,) studied this question.