基于光伏组件老化特性解耦的渔光互补光伏短期出力预测方法OA北大核心CSTPCD

The"fishery-solar hybrid project"effectively improves land use efficiency and reduces water evaporation.However,photovoltaic panels arranged on the water surface are affected by many factors,such as aging and power decline,making the prediction of water surface photovoltaic output challenging.The existing short-term photovoltaic output prediction technology extensively applies similar day selection to improve the prediction accuracy.However,it lacks consideration of aging phenomena,resulting in insufficient prediction accuracy for aquatic photovoltaic.Therefore,this paper proposes a short-term output prediction method for complementary fishing and solar power that decouples the aging characteristics of photovoltaic modules.First,an integrated learning model based on stacked recursive autoencoder-based learners was constructed to decouple the aging phenomenon of the dataset.Subsequently,the grey correlation analysis method was used to select similar days to train the deep learning predictor for the predicted time,and an improved Honey Badger algorithm was proposed to optimize the parameters of the base learners.Finally,the effectiveness and superiority of the proposed method were verified through a case study of the Fishery Light Complementary Project in Lufeng City,Yunnan Province.