摘要
风力发电功率受到风速、风向等气象因素而呈现强波动性与随机性,对超短期风电功率精确预测提出了挑战。针对上述问题,提出一种基于双重注意力机制-CNNGRU混合神经网络的超短期风电功率预测方法。首先,为提升模型对输入数据矩阵的时空特性提取能力,提出一种基于卷积神经网络结合门控循环单元神经网络的组合算法;其次,为提升模型对强相关性特征时序提取能力,提出一种特征注意力机制以自主分析历史信息与输入特征之间的关联关系;进一步,为提升模型预测效果的稳定性,建立一种时序注意力机制以自主选取CNNGRU网络强相关性时间点的历史信息;最后,得出最终预测结果。采用中国西北某集中式风电场数据进行算例验证,结果表明,所提方法能有效提高预测精度,具有一定的工程实用价值。
The power of wind generation shows strong volatility and randomness due to meteorological factors such as wind speed and wind direction,which poses a challenge to the precise prediction of ultra-short-term wind power generation.In order to solve the above problems,this paper proposes an ultra-short term wind power prediction method based on dual attention mechanism(AD)-CNNGRU hybrid neural network.First,in order to improve the ability of the model to extract the spatio-temporal characteristics of the input data matrix,the paper presents a combination algorithm based on convolutional neural network combined with gated recurrent unit neural network.Secondly,in order to improve the model's ability to extract time series of strongly correlated features,the paper proposes a feature attention mechanism to independently analyze the relationship between historical information and input features.Furthermore,in order to improve the stability of the model's prediction effect,the paper establishes a time-series attention mechanism to independently select historical information of CNNGRU network at strongly correlated time points.Finally,the paper obtains the final prediction results,and uses the data of a centralized wind farm in Northwest China to verify the results.The results show that the proposed method can effectively improve the prediction accuracy and has certain engineering practical value.
作者
王雅兰
田野
杨丽华
WANG Yalan;TIAN Ye;YANG Lihua(Marketing Service Center(Measurement Center),State Grid Hubei Electric Power Co.,Ltd.,Wuhan Hubei 430077,China;Electric Power Research Institute,State Grid Hubei Electric Power Co.,Ltd.,Wuhan Hubei 430077,China)
出处
《湖北电力》
2021年第3期23-28,共6页
Hubei Electric Power
基金
国家电网有限公司科技项目(项目编号:520626160052)。
关键词
风电功率
数据挖掘
超短期功率预测
混合神经网络
注意力机制
wind power
data mining
ultra-short-term power prediction
hybrid neural network
attention mechanism