Co-Occurrence Predictor for Wind Power Output

Abstract

AbstractWind power prediction has received extensive attention in recent years. In the open literature, problems such as imprecise statistical data and inaccurate prediction models still exist. To solve these problems, the authors first used k-means algorithm to cluster the measured data and initialized the centers by simulated annealing. Second, considering the co-occurrence information between wind power output and the relevant parameters, several matrices are introduced to calculate the co-occurrence number. Finally, a co-occurrence predictor was designed to calculate the wind power output by the approximate posterior probability. Experiments were conducted on a set of measured data. Experimental results suggest that this method outperforms the state of the art. Specifically, one only needs to calculate the co-occurrence tables, so the calculation of this model is less than any of the other wind power prediction model.