基于变分模式分解的混合风电预测研究

打开文本图片集

中图分类号：TP19；TP183 文献标识码：A 文章编号：2096-4706（2026）03-0169-09

Abstract： High-precision prediction of wind power generation is of great significance for the optimal scheduling and stableoperationofsmartgrids.However，windpowertime-seriesdataexhibitcharacteristicsofonlinearity，non-statiarity andlong-termdependene，makingitdificultfortraditionalmethodstoestablshefetivemodels.Fortisrason，ahbrid prediction model named VMD-BiLSTM is proposed based on Variational Mode Decomposition （VMD）and Bidirectional Long Short-Term Memory（BiLSTM）network.ThemodelusesVMDtodecompose theoriginalwind power data intomultiple stable modalcomponentstoallviatethecomplexcharacteristicsofthedata，andadoptsBiSTMtoextractthetemporaldependency information ineachcomponent toachieve acurate predictionof wind power trends.Experimentalresults ontwopublic wind power datasets show that VMD-BiLSTMoutperforms many other advanced models in terms of data fiting and prediction accuracy，and effectivelyreduces the impactofintermitent instabilty in wind power generationonpredictionresults.This research provides new ideas forthe modeling and predictionof complex wind power data and ofers strong data support for decision-making in smart grids.

Keywords： time series prediction; wind power; Big Data; Deep Learning; mode decomposition

0 引言

风电预测研究的背景和意义在于其对推动能源结构转型、提升风电利用效率和保障电网安全稳定运行的关键作用[]。（剩余10243字）