摘要
为提高风电功率预测的精确度,提出一种基于Kriging和长短期记忆网络的风电功率组合预测模型。首先,将风速、风向、空气密度、转速、偏航角和桨距角作为输入向量,并利用偏互信息理论对这些向量进行加权处理,建立基于Kriging的风电功率线性分量预测模型。然后,将预测出的线性分量加前述加权监测量作为输入,使用长短期记忆网络预测出功率的非线性分量。最后,将两者的预测结果相结合,得出风电功率的最终预测值。实例结果表明,该模型能够利用Kriging和长短期记忆网络的优势,预测性能指标得到提高。
In order to improve the accuracy of wind power prediction,a combined wind power prediction model based on Kriging and long short-term memory is proposed.Firstly,wind speed,wind direction,air density,rotor speed,yaw direction and pitch direction are taken as input vectors,and these vectors are weighted by the theory of partial mutual information,and a linear component prediction model of wind power based on Kriging is established.Then,the predicted linear component and the aforementioned weighted monitoring quantity are taken as the input,and the nonlinear component of power is predicted using the long short-term memory network.Finally,the final prediction value of wind power is obtained by combining the two prediction results.The example shows that the model can take advantage of Kriging and long short-term memory network,and improve the prediction performance.
作者
李俊卿
李秋佳
Li Junqing;Li Qiujia(School of Electrical and Electric Engineering,North China Electric Power University,Baoding 071000,China)
出处
《太阳能学报》
EI
CAS
CSCD
北大核心
2020年第11期241-247,共7页
Acta Energiae Solaris Sinica
基金
国家留学基金委资助项目(201706735002)
河北省自然科学基金(2014502015)。
关键词
风电功率
预测分析
数据处理
长短期记忆
KRIGING模型
偏互信息
wind power
predictive analytics
data processing
long short-term memory
Kriging model
partial mutual information
