基于混合分解和PCG-BiLSTM的风速短期预测  

SHORT-TERM PREDICTION OF WIND SPEED BASED ON HYBRID DECOMPOSITION AND PCG-BiLSTM

作　　者：毕贵红[1] 黄泽 赵四洪 谢旭 陈仕龙[1] 骆钊[1] Bi Guihong;Huang Ze;Zhao Sihong;Xie Xu;Chen Shilong;Luo Zhao(School of Electric Power Engineering,Kunming University of Science and Technology,Kunming 650500,China;Nuozhadu Hydropower Plant of Huaneng Lancang River Hydropower Inc,Puer 665000,China)

机构地区：[1]昆明理工大学电力工程学院,昆明650500 [2]华能澜沧江水电股份有限公司糯扎渡水电厂,普洱665000

出　　处：《太阳能学报》2024年第1期159-170,共12页Acta Energiae Solaris Sinica

基　　金：国家自然科学基金(51907084)。

摘　　要：为降低风速的随机性对风力发电的影响,提高风速短期预测的精准度,提出一种基于混合分解、双通道输入、多分支PCG-BiLSTM深度学习模型的短期风速预测方法。首先,将全年风速数据分为春、夏、秋、冬4个季度,选取春季作为主要实验对象;其次,利用奇异谱分解(SSD)和变分模态分解(VMD)以降低原始春季风速数据复杂度,生成具有不同模态且复杂度低的子分量,两种不同模式子分量组合为混合分量,实现不同模式分解算法的优势互补;最后,将混合分量以双通道的形式输入到多分支PCG-BiLSTM深度学习模型中,其模型的每个分支由卷积神经网络(CNN)与门控循环单元(GRU)并联组成时空特征提取模块,用于提取两种分解分量组合的混合分量的时空特征,各分支提取对应混合分量的时空特征经聚合后再由双向长短期记忆网络(BiLSTM)进一步提取风速信号的正向和反向双向波动规律,进而得到最终的风速预测结果。多组实验结果表明:提出的组合预测方法在短期风速预测中具有较高的精度和泛化能力,优于其他传统预测方法。In order to reduce the impact of the randomness of wind speed on wind power generation and improve the accuracy of wind speed short-term prediction.In this paper,a short-term wind speed prediction method based on hybrid decomposition,two-channel input and multi-branch PCG-BiLSTM deep learning model is proposed.Firstly,the annual wind speed data were divided into four seasons:spring,summer,autumn and winter,and spring was selected as the main experimental subjects.Secondly,singular spectrum decomposition(SSD)and variational modal decomposition(VMD)are used to reduce the complexity of the original spring wind speed data,generate subcomponents with different modalities and low complexity,and the two different modal subcomponents are combined into a mixed component to achieve the complementarity of the advantages of different mode decomposition algorithms.Finally,the mixed components are input to the multi-branch PCG-BiLSTM deep learning model in the form of two channels,and each branch of the model consists of a parallel convolutional neural network(CNN)and a gated recurrent unit(GRU)to obtained a spatio-temporal feature extraction module for extracting the spatio-temporal features of the mixed components of the combination of two decomposition components,and the spatio-temporal features of the corresponding mixed components extracted by each branch are aggregated and then further extracted by a bidirectional long short-term memory network(BiLSTM)for the forward and reverse bi-directional fluctuation regularity of the wind speed signals,and then the final wind speed prediction results are obtained.Several sets of experimental results show that the combined prediction method proposed in this paper has high accuracy and generalization ability in short-term wind speed prediction,which is better than other traditional prediction methods.

关 键 词：风速 预测 深度学习 混合分解 并联网络 

分 类 号：TM614[电气工程—电力系统及自动化]