Wind power 24-h ahead forecast by an artificial neural network and an hybrid model: Comparison of the predictive performance