Fuzzified Viterbi algorithm for hour-ahead wind power prediction

This paper presents a new fuzzy stochastic method for very short-term (1 hour) wind prediction to address both the stochastic and linguistic uncertainties of wind power prediction in electrical power systems.. Past wind farm power production data are required to develop a hidden Markov model (HMM) of the power network. The transition probabilities of the HMM are estimated using a fuzzy stochastic approach that improves the quality of the estimates. The fuzzy estimation can use a variety of membership functions and the effect of the choice of membership function on the estimation is investigated by comparing the results for interval and triangular membership functions. State prediction is achieved using a fuzzy Viterbi algorithm (VA) derived using the extension principle. Computer simulations using Northwestern weather recordings from the Bonneville Power Administration (BPA) website show good correlation between our predictions and the actual data.