Configuration of a Wind Power Forecasting Model based on Fuzzy c-Means Clustering

Stéphanie Monjoly and Ruddy Blonbou


Fuzzy c-Means, Intensity of turbulence, Bayesian neural networks, Wind power forecasting


In this paper we present a method intended to estimate the parameters of a prediction model based on a Bayesian neural network. The method first performed a classification of a set of 12 hours long pre-recorded wind power sequences, based on the level of their intensity of turbulence, using the Fuzzy c-Means method. Then using a Markov chain like approach, it determines the matrix of transition that gives the transition probability from the current wind power regime of turbulence defined from the most recent 12 hours recorded wind power data, to the next regime of turbulence of the wind power during the incoming hour. Finally, the classification and the matrix of transition allows us to propose an adaptive forecasting scheme that calculates the values of the prediction scheme’s time scales parameters which are conditioned by the nature of the transition combinations proposed by the matrix of transition.

Important Links:

Go Back