The economic and climate contexts require to use more electricity from wind farms. However this kind of production is intermittent, therefore it is necessary to forecast this resource at least 1 day ahead. Our laboratory has developed a forecasting model of wind-based electricity generation based on a global meteorological model (GFS) with a resolution of 50 km and 3 h. But this model has a resolution too coarse for a wind farm. So we have configured the regional model WRF with resolution of 2 km and 15 min to obtain better forecasts. Finally, the WRF model provides better forecasts, but both must be adjusted to take into account the direct environment of the wind farm.