Accurate prediction of wind speed along high-speed rail lines is a fundamental requirement for railway disaster warning systems. To enhance the capability to respond to and handle sudden events caused by strong winds, a short-term wind speed prediction method based on the subtraction average based optimizer (SABO) algorithm optimized long short-term memory (LSTM) neural network was proposed. Firstly, considering the nonlinearity and non-stationarity of wind speed, the min-max (MM) method was used to normalize the wind speed data. Secondly, the "- v" method in the SABO algorithm was employed to search and optimize the key parameters of the LSTM model, constructing a wind speed prediction model. Finally, the effectiveness of the model was tested using measured wind speed data collected from wind speed collection points along the Baoji-Lanzhou high-speed railway in China. Experimental results show that the SABO algorithm's optimization effect is better, and the prediction accuracy is higher. The average absolute error (MAE), mean absolute percentage error (MAPE), and root mean square error (RMSE) of the constructed model are 11.96%, 1.23%, and 16.47%, respectively, with a coefficient of determination (R²) of 0.995. Compared to other models, the LSTM neural network optimized by the SABO algorithm exhibits better fitting performance and higher prediction accuracy in short-term wind speed prediction, providing a new method and approach for wind prediction and warning along high-speed railway.