One of the problems in using wind energy is that the electricity generated by wind is more unstable than electricity generated by other sources, and as a result, integrating the use of wind energy with other power generation systems is a challenge. Reduced the problem by having accurate information about the mean and wind speed volatility. In this article, we model the mean and wind speed volatility. The data under study is the average weekly wind speed from the beginning of 2002 to the end of 2016 at the Rasht meteorological station. To model, we first fit a moving average autoregressive model to the average wind speed, then use the GARCH models for the remainder of the model fit. Finally, using the Bayesian criterion, we select the best model for modeling and conclude that ARMA-GARCH models are suitable for wind speed modeling.