Wind is a renewable, sustainable, and clean source of energy. This has led to wind gaining a lot of attention in recent decades as a reliable alternative to fossil fuels. However, wind speed fluctuations complicate its integration with power grids. To tackle this issue, this paper proposes a new wind speed prediction model that combines four techniques: Discrete Wavelet Transform, which smooths the wind speed signal; Mutual Information, which selects the most informative part of the wind speed time series; Coot Optimization Algorithm for optimal feature selection; and Bidirectional Long Short-Term Memory for capturing complex patterns. To evaluate the efficiency of the proposed model, its performance was measured using error metrics such as mean squared error, mean absolute error, mean absolute percentage error, coefficient of determination ([Formula: see text]), and median absolute error. The proposed model was examined using two different wind speed datasets and achieved high prediction accuracy. Additionally, 14 different benchmark models were created, and their prediction results were compared with those of the proposed model. A comparison between the results of the proposed model and benchmark models demonstrated the superiority of the proposed model.