SHORT-TERM PREDICTION OF REGIONAL WIND-SOLAR POWER OF EEMD-RL-GWO-LSTM ON REVERSE CLOUD GREY CORRELATION SIMILAR DAYS
Aiming at the problems of incomplete consideration of meteorological factors and non-consideration of wind-solar power correlation in wind-solar power prediction by existing methods,a short-term prediction method of wind-solar power is proposed.Firstly,the cloud model is used to characterize the uncertainty of wind-solar output,and the influence of different meteorological characteristics on the output power is analyzed by the reverse cloud combined with the grey correlation degree,and the selection criteria and comprehensive scoring index are set up.Secondly,the power data of similar days are decomposed into subsequences by ensemble empirical mode decomposition(EEMD);Finally,the sub-sequences and meteorological data are trained as the forecast inputs of the improved long and short term memory network(LSTM)model optimized by the grey wolf algorithm(GWO)based on refraction learning strategy(RL).The sub-sequences of the forecast days are predicted separately and superimposed to obtain the prediction of the short-term regional wind-solar power.The designed model is verified by the data of a certain wind-solar farm located in northwest China.The experimental results show that,compared with the existing prediction models,the proposed method takes into account the weather factor,has high prediction accuracy,and can better provide a reference for the power prediction of regional wind-scenic combined farms.
reverse cloud grey correlation similar daysensemble empirical mode decomposition(EEMD)RL-GWO-LSTM neural networkshort-term wind-solar power prediction
万方数据
维普
