摘要
针对风电数据存在维度多、波动大等特点而加大风电功率预测难度的问题,本文提出一种基于卷积神经网络(convolutional neural networks,CNN)和双向门控循环单元(bidirectional gated recurrent unit,BiGRU)的风电功率预测模型。该模型通过Pearson相关系数筛选最佳的历史功率和气象因素组合,使用CNN网络提取原始数据的时序特征,然后利用BiGRU网络捕捉这些特征之间的时序依赖关系,最终得到风功率预测值。算例分析表明,本文所提CNN-BiGRU模型比传统的BP和BiGRU神经网络模型具有更高的预测精度。
Due to the characteristics of multi-dimensional and largefluctuation of wind power data,it is difficult to predict wind power,this paper proposes a wind power prediction model based on convolutional neural networks(CNN)and bidirectional gated recurrent unit(BiGRU).The best combination of historical power and meteorological factors is selected by Pearson correlation coefficient in this model,the CNN network is used to extract the time series features of the original data,and then the BiGRU network is used to capture the time series dependence between these features,andfinally,the wind power prediction values are obtained.The analysis of calculation examples shows that the CNN-BiGRU model proposed in this paper has higher prediction accuracy than the traditional BP and BiGRU neural network models.
作者
万黎升
陈凡
傅裕
井思桐
WAN LiSheng;CHEN Fan;FU Yu;JING Sitong(PowerChina Jiangxi Electric Power Engineering Co.,Ltd.,Nanchang 330096,China)
出处
《电力勘测设计》
2024年第7期23-28,57,共7页
Electric Power Survey & Design
关键词
风电功率预测
Pearson相关系数
卷积神经网络
双向门控循环单元
wind power prediction
pearson correlation coefficient
convolutional neural network
bidirectional gated recurrent unit
