Modeling box - jenkins applied a prediction of wind speed in the regions of the brazilian northeast for the fins of wind generation

Abstract

In the present work, a study was carried out to predict monthly average wind speed in regions of the Brazilian Northeast. For this purpose, the Box - Jenkins modeling methodology was applied to the 10 m high wind speed data from January 2010 to December 2013. The forecast in all study locations was for the year 2014 through of the SARIMA model, which predominated in practically all regions, that is, an indication that the ideal forecast model must necessarily be introduced to the seasonal component. The prediction was efficient in some regions, for example, in Aracaju it was possible to find a MAPE error of 4.66%. In the localities of Aracaju and Salvador it is possible to identify that the predicted series tend to have similar behavior to the observed series regarding the similarity of maximum and minimum wind speed. This work could be used as a wind speed prediction tool to study and advance wind generation in several regions, providing decision makers with local wind exploitation, since it will be possible to estimate the wind regime in the future.