现代电子技术2024，Vol.47Issue(18)：1-7,7.DOI:10.16652/j.issn.1004-373x.2024.18.001
基于信息调控和MATCN的超短期风电功率多步预测
Ultra-short-term wind power multi-step forecasting based on information regulation and MATCN
摘要
Abstract
The effective forecasting of fluctuating wind power is an important guarantee for the balance of power supply and demand and the stable operation of the system.Therefore,a method of ultra-short-term wind power multi-step forecasting based on information regulation and multi-head attention temporal convolution network(MATCN)is proposed.High-order items and interactive items are derived from the existing data to increase the proportion of the number of feature sequences and effective features.The variational mode decomposition(VMD)is used to split the complex wind power data structure,the important sequence components are retained according to the calculation results of sub-sequence correlation and variance contribution rate,and other components are aggregated to reduce the calculation burden,and shorter training time.attention mechanism is introduced to construct the MATCN,and the transmitted information between convolution units within the network is adjusted by the attention score,so as to realize the prediction of each sequence component of the model.The sequence component prediction values are reconstructed to obtain the final output result.The proposed model is compared and verified on example data,and the results show that the model has excellent prediction effects under different strides.关键词
风电功率/多步预测/变分模态分解/多头注意力时间卷积网络/注意力机制/信息调控Key words
wind power/multi-step forecasting/variational mode decomposition/multi-head attention temporal convolution network/attention mechanism/information regulation分类
电子信息工程引用本文复制引用
陈磊,黄凯阳,张怡,陈禹,张志瑞,尹振楠..基于信息调控和MATCN的超短期风电功率多步预测[J].现代电子技术,2024,47(18):1-7,7.基金项目
国家重点研发计划项目(2021YFE0190900) （2021YFE0190900）
教育部产学合作协同育人项目(230802495182120) （230802495182120）
华北理工大学研究生教育教学改革项目(YJG202308) （YJG202308）
