Forecasting meteorology models has become a main tool, in order to obtain energy exploitation, in turn eco- logically. Currently Artificial Intelligence (AI) has changed the approach to solve the problems of prediction, among others. The change in approach is due to the AI that has modern learning techniques with mathematical models closer to reality. Our approach is to implement feed-forward neural networks architecture into wind forecasting models. The objective of our work is to predict the mean monthly wind speed behavior for year 2012 and beyond, as close to reality in the city of San Jose de Cucuta-Colombia.