基于SFLA和MSISSA-ANFIS的超短期光伏功率动态预测方法  

ULTRA-SHORT-TERM PV POWER DYNAMIC PREDICTION METHOD BASED ON SFLA AND MSISSA-ANFIS

作　　者：李练兵[1] 高国强 陶鹏 张超 赵莎莎 陈伟光 Li Lianbing;Gao Guoqiang;Tao Peng;Zhang Chao;Zhao Shasha;Chen Weiguang(State Key Laboratory of Electrical Equipment Reliability and Intelligence(Hebei University of Technology),Tianjin 300131,China;School of Artificial Intelligence and Data Science,Hebei University of Technology,Tianjin 300131,China;State Grid Hebei Electric Power Co.,Ltd.Marketing Service Center,Shijiazhuang 050035,China)

机构地区：[1]省部共建电工装备可靠性与智能化国家重点实验室(河北工业大学),天津300131 [2]河北工业大学人工智能与数据科学学院,天津300131 [3]国网河北省电力有限公司营销服务中心,石家庄050035

出　　处：《太阳能学报》2024年第10期326-335,共10页Acta Energiae Solaris Sinica

基　　金：河北省省级科技计划(20314301D)。

摘　　要：为进一步提高光伏功率预测的精度,提出一种基于SFLA、MSISSA和ANFIS的超短期光伏功率日内动态预测模型。首先针对ANFIS模型受成员函数影响较大的缺点采用MSISSA对其进行优化,并结合SFLA选取相似日的方法,构建基于SFLA和MSISSA-ANFIS的功率预测模型。然后根据相关性较高的功率、气象特征与相似日集合构建特征向量对未来4 h的光伏功率进行预测。最后将从小型气象站获得的实时更新的未来气象数据存入数据库,每隔15 min预测一次,实现光伏功率的日内动态预测。结果表明所提方法提高了超短期光伏预测的精度。PV power output is characterized by volatility and randomness,and accurate power prediction has important application value for safe grid operation in the process of achieving grid connection.In this paper,an intra-day dynamic prediction model of ultra-short-term PV power based on SFLA,MSISSA and ANFIS is proposed.Firstly,MSISSA is used to optimize the ANFIS model for its drawback of being influenced by the membership function,and the power prediction model based on SFLA and MSISSA-ANFIS is constructed by combining the SFLA method of selecting similar days.Then a feature vector is constructed to predict the PV power for the next 4 h based on the set of power,meteorological features and similar days with high correlation.Finally,the real-time updated future meteorological data obtained from small weather stations are stored in the database and predicted every 15 min to realize the intra-day dynamic prediction of PV power.The results show that the proposed method improves the accuracy of ultra-short-term PV prediction。

关 键 词：光伏功率预测 时间序列 自适应神经模糊推理系统 算法优化 相似日选取 

分 类 号：TM615[电气工程—电力系统及自动化]