SVMD-PE-BP-Transformer短期光伏功率预测
SVMD-PE-BP-Transformer Short-Term PV Power Prediction
王瑞 1靳鑫鑫 1逯静1
作者信息
- 1. 河南理工大学计算机科学与技术学院,河南焦作 454000
- 折叠
摘要
考虑到光伏功率受气象因素变化影响而波动性大难以预测的问题,将逐次变分模态分解SVMD-排列熵PE与BP-Transformer相结合,给出了一种组合预测方法,以下简称SPBT模型.在去除非相关因子的基础上,利用SOM聚类方法,对全年光伏数据进行 3 种类型的分类;针对光伏发电初始时序中所蕴含的重要信息,利用SVMD自适应K值的方法,对其进行分解.再利用PE方法计算各个子序列的熵值,即序列的起伏复杂程度,根据熵的大小,对频率接近的成分进行重构,将其分为两个区间:复杂度低的部分和复杂度高的部分.最后利用BP网络与Transformer分别对其进行预测,并对预测输出进行综合处理.该文以江苏省一光伏电站观测的气象与功率数据为例,通过比较试验验证了该模型的优势,该模型具有较低的预测误差,有助于提高预测精度.
Abstract
Considering the question of high volatility and difficulty in predicting photovoltaic power due to changes of meteorological factors,a combined prediction method,hereinafter referred to as the SPBT model,is given by combining the successive variational modal decomposition SVMD-arranged entropy PE with the BP-Transformer.First,on the basis of removing non-correlated factors,the SOM clustering method is utilized to classify the year-round PV data into three types;second,the SVMD adaptive K-value method is used to decompose the PV power generation for the important information embedded in the initial time series.In addition,the PE method is used to calculate the entropy value of each subsequence,i.e.,the complexity of the ups and downs of the sequence,and according to the magnitude of the entropy,the components with close frequency are reconstructed and divided into two intervals:the part of low complexity and the part of high complexity,and finally the BP network and Transformer are used to make predictions respectively,and the prediction outputs are processed comprehensively.The paper takes the meteorological and power data observed in a PV power station in Jiangsu Province as an example,and verifies the advantages of the model through comparative tests,suggesting that the model is of low prediction error and helps to improve the prediction accuracy.
关键词
逐次变分模态分解/排列熵/Transformer/功率预测Key words
SVMD/PE/Transformer/power prediction引用本文复制引用
出版年
2024
NETL
NSTL
万方数据