MULTIPLE OBJECTIVE INTERVAL PREDICTION OF LSTM WIND POWER BASED ON COMPETITIVE LEARNING MECHANISM
In order to further improve the comprehensive effect of interval prediction,a multiple objective interval prediction method of LSTM wind power based on competitive learning mechanism is proposed.The upper and lower bounds estimation model based on LSTM was proposed to construct the multiple objective prediction model of wind power interval,and the relationship between the estimation error and the average width of prediction interval in the multiple objective system was studied.Further considering the prediction error,a new partial least squares evaluation index was introduced.In addition,by introducing competitive learning mechanism,an improved non dominated quick sort genetic algorithm was proposed,which effectively realized multiple objective optimization.Two real wind power data sets were used to verify the proposed method.The results show that the proposed method has high prediction accuracy.
Wind power forecastingLong and short-term memory networkInterval forecastingGenetic algorithm
国家科技期刊平台
NSTL
万方数据
