Short-term wind power data forecasting by particle swarm optimization dynamic GM(1,1) model

A particle swarm optimization (PSO) dynamic GM(1,1) model is proposed in addressing problems of poor information, low accuracy and high uncertainty in solving the short-term wind power data forecasting. The background value of the model is optimized by PSO with different fitness with other researches, the forecasting data of which can be continuously smoothed. Furthermore, a self-adaptive intelligence model is introduced in proposed model to deal with the homogeneous and non-homogeneous exponent functions to improve the forecasting accuracy. The residual model is also added to the proposed model to deduced the uncertainty brought from surroundings. The experience validates the proposed model is effective in short-term wind power output forecasting in a Belgium wind farm.