基于ICEEMDAN-PE-GDBO-LSSVM的风电功率预测

打开文本图片集

中图分类号：TN911.7-34；TM614 文献标识码：A 文章编号：1004-373X（2025）10-0057-06

Abstract：Withthehighpenetrationofrenewableenergy，especiallywind power，thepowergrid isfacingunprecedented chalengesofuncertaintyandvolatity.Inordertoacuratelypredictwindpower，acombinedmodelbasedonimproved completeensemble empirical mode decompostionwith adaptivenoise （ICEEMDAN）-permutation entropy（PE）-improved dung beetle optimizationalgorithm（GDBO）-least support squaresvector machine （LSSVM） isproposed.ICEEMDANis used to decompose thewindpowerdata toreducethecomplexity.Thecomponentsobtainedafterdecompositionareagregatedaccording toPE，andthenthekeyparametersofLSSVMareoptimizedbymeansofGDBOalgorithm toobtain thebestpredictionmodel. Theoptimizationmodelisusedtopredictandsuperimposetheaggregationcomponents toobtain the totalpredictionresult.The experimentalverification isconductedbasedonthedomesticwindfarmdataset.Ttheresultsshowthattheproposed method has high predictionaccuracyandtheroot meansquare errris 61.39%lowerthan thatofthesingle LSSVMmodel，whichhasa broader application prospect in engineering practice.

Keywords：wind power prediction; ICEEMDAN; GDBO; PE;improved complete ensemble empirical mode decomposition; LSSVM; component polymerization

0 引言

随着社会对环境保护和可持续发展的重视程度不断提升，清洁能源产业正经历前所未有的蓬勃发展。（剩余6339字）