A large, fully steerable radio telescope is susceptible to the wind load, leading to structure deformation and pointing deviation of the telescope. To effectively suppress the influence of dynamic wind load, the wind resistance control of the telescope is carried out based on wind speed forecasting. This study developed a wind speed forecasting model to efficiently forecast the wind speed at the telescope position. The proposed model successfully eliminates the random noise of the original wind speed, effectively extracts the wind speed features and solves the automatic optimization of the hyperparameters of the forecasting network. This model significantly improves the accuracy and reliability of wind speed forecasting. To verify the forecasting performance of the proposed model, the wind data from the Qitai Radio Telescope site is examined as a case study. The wind speed forecasting model's MAE, RMSE and MAPE are 0.0361, 0.0703 and 3.87%, respectively. The performance of the proposed model meets the requirements of wind resistance control and can provide data support for the radio telescope.