One day ahead prediction of wind speed class

This paper deals with the problem of clustering daily wind speed time series based on two features referred to as Wr and H, representing a measure of the relative daily average wind speed and the Hurst exponent, respectively. Daily values of the pairs (Wr, H) are first classified by means of the fuzzy c-means unsupervised clustering algorithm and then results are used to train a supervised MLP neural network classifier. It is shown that associating to a true wind speed time series a time series of classes, allows performing some useful statistics. Further, the problem of predicting 1-step ahead the class of daily wind speed is addressed by introducing NAR sigmoidal neural models into the classification process. The performance of the prediction model is finally assessed.