摘要
由于风力发电的随机性和间歇性,风功率预测不仅需要准确的点预测,而且需要可靠的区间预测和概率预测来量化风功率的不确定性。提出了一种基于变分模态分解(variational mode decomposition,VMD)和分位数卷积-循环神经网络的风功率概率预测模型。首先,使用VMD技术将原始风功率数据序列分解为一系列特征互异的模态分量,再通过卷积神经网络(convolutional neural network,CNN)提取反映各模态分量动态变化的高阶特征。然后,基于提取的高阶特征进行分位数回归建模,采用长短期记忆(long shortterm memory,LSTM)循环神经网络预测未来任意时刻不同分位数条件下的风功率值。最后,利用核密度估计(kernel density estimation,KDE)得到风功率概率密度曲线。以中国某风电场数据作为算例测试,证明了所提出模型的有效性。
Due to the randomness and intermittency of wind power,wind power forecasting requires not only accurate point forecasting,but also reliable interval and probabilistic forecasting to quantify the uncertainty of wind power.This paper proposes a probabilistic wind power forecasting method based on variational mode decomposition(VMD)and quantile convolution-recurrent neural network.Firstly,this method uses VMD to decompose the original wind power sequence into a series of modal components with different characteristics.Secondly,the convolutional neural network(CNN)is used to extract high-order features reflecting the dynamic changes of each modal component.Then,the quantile regression is performed by the long short-term memory(LSTM)recurrent neural network based on the high-order features to obtain the predicted values for different quantiles.Finally,the kernel density estimation(KDE)is employed to estimate the probability density curve of wind power.The effectiveness of the proposed method is verified with the example test using datasets from the wind farm in China.
作者
沙骏
徐雨森
刘冲冲
冯定东
胥峥
臧海祥
SHA Jun;XU Yusen;LIU Chongchong;FENG Dingdong;XU Zheng;ZANG Haixiang(Yancheng Power Supply Branch of State Grid Jiangsu Electric Power Co.,Ltd.,Yancheng 224008,China;College of Energy and Electrical Engineering,Hohai University,Nanjing 210098,China)
出处
《中国电力》
CSCD
北大核心
2022年第12期61-68,共8页
Electric Power
基金
国家自然科学基金资助项目(52077062)
国网江苏省电力有限公司科技项目(J2020122)。
关键词
风功率预测
变分模态分解
卷积神经网络
长短期记忆循环神经网络
分位数回归
wind power prediction
variational mode decomposition
convolutional neural network
long short-term memory recurrent neural network
quantile regression
