Wind power forecasting based on a Markov chain model of variation

In order to provide more decision-making information and improve the effects of probabilistic forecasting, a novel wind power probabilistic forecasting approach based on Markov chain model of variation is proposed in this paper. The transition probabilities matrix of Markov chain model is built from the variations of historical wind power data to compute probability distributions and deterministic wind generation at future moments. Refined state space is built because the value of variation is much smaller than wind generation, which could improve the prediction accuracy. The effectiveness and higher accuracy of proposed approach are proved by actual data from wind farm.