The comprehensive evaluation and forecasting of wind speed are paramount to operational efficiency and planning within sectors such as renewable energy, weather prediction, aviation, and maritime navigation. This study examines the potential for improving the accuracy of wind speed predictions by integrating the ReliefF algorithm with Long Short-term Memory (LSTM) networks. The results of this approach are compared with the traditional ANN model using the performance metrics such as MSE (Mean Squared Error), MAE (Mean Absolute Error), and RMSE (Root Mean Squared Error). The ReliefF algorithm is employed to select relevant features from meteorological data, which improves the efficiency and performance of the LSTM model. The LSTM network, known for its capability to capture temporal dependencies in time-series data, is then trained on the selected features. The proposed approach is trained using a real-world wind speed dataset, namely the Jafrabad Coastal Dataset, demonstrating significant improvements in forecasting accuracy compared to traditional ANN networks. The combination of feature selection through ReliefF and the advanced sequence modeling of LSTM provides a robust framework for precise wind speed prediction, which can be highly beneficial for enhancing operational decisions in wind energy systems and other weather-dependent sectors.