机器学习在光伏发电功率预测中的应用分析
Application of Machine Learning in Photovoltaic Power Generation Prediction
李特 1王尧 1武文起 2赵炜1
作者信息
- 1. 国网河北省电力有限公司信息通信分公司,河北 石家庄 050000
- 2. 河北水利电力学院,河北 沧州 061001
- 折叠
摘要
光伏发电存在典型的间歇性、随机性、不确定性特征,准确的光伏发电功率预测模型对于保障电力平衡、优化方式安排、促进新能源消纳十分必要.机器学习通过学习已有的大量数据形成预测及判断,能够挖掘数据内在价值,在光伏发电领域取得丰富成果.给出基于机器学习的光伏发电功率预测框架,重点分析了传统机器学习模型、深度学习模型、组合集成学习模型、在线学习模型、物理数据联合驱动模型5种光伏发电预测方法,并对每种方法给出研究建议.
Abstract
Photovoltaic power generation has typical characteristics of intermittency,randomness,and uncertainty.An accurate photovoltaic power generation prediction model is essential for ensuring power balance,optimizing mode arrangements,and pro-moting new energy consumption.Machine learning can form predictions and judgments by learning a large amount of existing data,and can explore the intrinsic value of the data,achieving rich results in the field of photovoltaic power generation.This arti-cle provides a framework for predicting photovoltaic power generation based on machine learning,with a focus on analyzing five types of photovoltaic power generation prediction methods:traditional machine learning models,deep learning models,combina-torial ensemble learning models,online learning models,and physical data joint driving models.And research recommendations are provided for each type of method.
关键词
机器学习/光伏发电预测/小样本学习/在线学习/集成学习Key words
machine learning/photovoltaic power generation prediction/small sample learning/online learning/integrated learning引用本文复制引用
出版年
2024
NETL
NSTL
万方数据