基于VMD-TCN-LSTM模型的短期光伏功率预测
Short-Term Photovoltaic Power Prediction Based on VMD-TCN-LSTM Model
何铖 1余成波 1龙铖 1朱春霖1
作者信息
- 1. 重庆理工大学电气与电子工程学院,重庆 400054
- 折叠
摘要
针对光伏发电易受气象因素影响导致发电功率不稳定问题,提出了一种基于变分模态分解(Variational Modal Decom-position,VMD)、时域卷积网络(Temporal Convolutional Network,TCN)和长短期记忆神经网络(Long Short Term Memory Net-work,LSTM)的短期光伏功率预测模型,以准确预测其发电功率,给电力系统的调度规划提供参考.上述模型首先使用VMD对光伏发电功率数据进行分解,得到多个不同频率的分量,再将光伏发电功率数据、分量数据以及气象数据输入TCN-LSTM,以对数据特征进行深度挖掘,经多次迭代最终输出预测结果.通过实验验证,上述模型的均方误差、平均绝对误差等指标有所提升,具有更高的预测精度.
Abstract
To address the problem that photovoltaic power generation is susceptible to meteorological factors that lead to unstable photovoltaic power generation,this paper proposes a short-term photovoltaic power prediction model based on Variational Modal Decomposition(VMD),Temporal Convolutional Network(TCN)and Long Short Term Memory Network(LSTM)to accurately predict the power generation and provide a reference for power system dispatch planning.The model firstly decomposes the photovoltaic power data using VMD to obtain several components with different frequencies,and then inputs the photovoltaic power data,component data and meteorological data into TCN-LSTM for deep mining of data features,and finally outputs the prediction results after several iterations.Through experimental verification,the model has improved the mean square error,mean absolute error and other indicators,and has higher prediction accuracy.
关键词
气象因素/光伏发电/功率预测/变分模态分解/时域卷积网络/长短期记忆神经网络Key words
Meteorological factors/Photovoltaic power generation/Predict the power generation/Variational modal decomposition/Temporal convolutional network/Long short-term memory network引用本文复制引用
基金项目
国家自然基金资助项目(61976030)
高端外国专家项目(GDW20165200063)
重庆市高校优秀成果转化项目(KJZH4213)
出版年
2024
NETL
NSTL
万方数据