The use of wind speed data is of undeniable interest in the operation of wind farms. As such, the techniques of artificial generation of wind data are used in case the need to have a large database is required. Several methods are used for this purpose. The most classic are those of normal and Weibull distributions. In this study, a new approach based on wavelet transformation is introduced. It is compared to classical methods. The results show that the wavelet method is an adequate alternative to conventional methods.