基于VMD-SE的电力负荷分量的多特征短期预测
Multi-feature Short-term Prediction of Power Load Components Based on VMD-SE
邵必林 1纪丹阳1
作者信息
- 1. 西安建筑科技大学管理学院,陕西西安 710399
- 折叠
摘要
为提高电力负荷的预测精度,提出一种基于VMD-SE的电力负荷分量的多特征短期预测方法.首先采用变分模态分解(VMD)将原始负荷分解为一系列模态分量与残差,VMD的分解层数由样本熵值(sample entropy,SE)确定;然后对比原始负荷与模态分量的SE值,重构为平稳分量和波动分量,来降低运算规模;同时利用皮尔逊相关系数来筛选特征变量,删除特征冗余,建立灰狼算法优化后的支持向量回归模型(GWO-SVR)和长短期记忆神经网络(LSTM)分别对平稳分量和波动分量预测;最后以某地区2018-2020 年用电负荷为例进行实验.实验证明:此模型精准度高达 94.7%,平均绝对百分误差降低到2.98%,具有更好的精准性和适用性.
Abstract
To improve the accuracy of power load prediction,a multi-feature short-term prediction method based on VMD-SE for power load components is proposed.Firstly,the variational modal decomposition(VMD)is used to decompose the original load into a series of modal components and residuals,and the decomposition level of VMD is determined by sample entropy(SE).Then the SE values of the original load and modal components are compared,and the original load series are reconstructed into stationary and fluctuating components to reduce the computational scale.At the same time,the Pearson correlation coefficient is used to screen feature variables and delete feature redundancy,and a support vector regression model(GWO-SVR)optimized by gray wolf algorithm and a long short term memory neural network are established to predict the stationary component and fluctuation component respectively.Finally,an experiment was conducted using the electricity load of an area in Xi'an from 2018 to 2020 as an example.The experiment proves that the accuracy of this model is as high as 94.7%,and the MAPE error is reduced to 2.98%,indicating good accuracy and applicability.
关键词
短期预测/VMD/样本熵/波动分量/平稳分量/GWO-SVR/长短期记忆神经网络Key words
short-term prediction/variational modal decomposition/sample entropy/fluctuation component/stationary component/GWO-SVR/short and long term memory neural network引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据