短期风电功率预测的深度学习模型  

Deep learning model for short-term wind power prediction

作　　者：辛征[1] 王琦 刘兴然 Xin Zheng;Wang Qi;Liu Xingran(School of Information and Electrical Engineering,Shandong Jianzhu University,Jinan,Shandong 250101,China)

机构地区：[1]山东建筑大学信息与电气工程学院,山东济南250101

出　　处：《计算机时代》2023年第2期33-36,41,共5页Computer Era

摘　　要：风电固有的间歇性和不确定性给准确预测风电出力带来了挑战,也给风电并网带来了棘手的问题。本文提出一种用于风电短期预测的卷积神经网络-双向长短期记忆网络(CNN-BiLSTM)混合深度学习模型。首先利用双向长短期记忆网络挖掘双向时间特征,然后利用卷积神经网络提取空间特征。最后使用本文模型对某风电场风电功率进行预测。对比结果表明,使用CNN-BiLSTM方法能显著改善预测性能,降低风电功率预测误差。The inherent intermittency and uncertainty of wind power pose challenges to accurately forecasting wind power output,as well as thorny issues for wind power grid integration. In this paper, a convolutional neural network-bidirectional long short-term memory network(CNN-BiLSTM) hybrid deep learning model is proposed for wind power short-term prediction. First, bidirectional temporal features are mined using BiLSTM, and then spatial features are extracted using CNN. Finally, the proposed method is used to predict the wind power of a wind farm. The comparison results show that using the CNN-BiLSTM method can significantly improve the prediction performance and reduce the wind power prediction error.

关 键 词：风电功率 短期预测 双向长短期记忆网络 特征提取 

分 类 号：TM732[电气工程—电力系统及自动化]