ANN approach to WECS power forecast

Abstract

In this work-in-progress the problem with the future integration of large quantity of wind generators in the Portuguese electric grid is presented. A method based in artificial neural networks (ANN) is used to predict the average hourly wind speed. The work starts by choosing the patterns set length, the ANN structure and the learning method. As well as the dimensions of the data sets, training, validation and test. The ANN is tested with several structures until it archives an acceptable ANN based model. The obtained model is used to predict the wind speed and to forecast the power produced. The results archived are discussed. The future work perspectives are present