第60卷第6期2023年6月15日电测与仪表ElectricalMeasurement&InstrumentationVol.60No.6Jun.15,2023基于Stacking集成学习的有源台区线损率评估方法董美娜"，刘丽平”,王泽忠"，王守强"，张子岩²，邹运（1.华北电力大学电气与电子工程学院，北京102206；2.中国电力科学研究院，北京100192）摘要：人工智能及机器学习的发展,为有源台区线损率的评估提供了薪新的思路。提出一种基于Stacking集成学习的有源台区线损率评估方法。从特定系统中提取有源台区数据,采用互信息等方法处理数据中异常值,并建立电气特征指标体系。考虑传统的机器学习与不同思想的集成学习算法之间的差异，综合线性模型与非线性模型，选择线性回归算法、随机森林算法、GBDT算法作为基学习器，构建多算法融合的Stacking集成学习模型。以某省有源台区数据为例,验证了所提方法的准确性和有效性。关键词：有源台区；线损率；互信息；集成学习；多算法融合D0I:10.19753/j.issn1001-1390.2023.06.019中图分类号：TM764AlinelossrateevaluationmethodbasedonstackingensemblelearningDongMeina',LiuLiping",WangZezhong',WangShouqiang',ZhangZiyan’,ZouYun?(1.SchoolofElectricalandElectronicEngineering,NorthChinaElectricPowerUniversity,Beijing102206,China.2.ChinaElectricPowerResearchInstitute,Beijing100192,China)Abstract:ThedevelopmentofartificialintelligenceandmachinelearningprovidedanewideafortheevaluationoflinelossrateoftransformerdistrictwithDG.AlinelossrateevaluationmethodbasedonStackingensemblelearningfortrans-formerdistrictwithDGwasproposedinthispaper.DataoftransformerdistrictswithDGwasextractedfromspecificsys-temsandtheoutliersinthedatawereprocessedbymeansofmutualinformationtoestablishtheelectricalcharacteristicin-dicatorsystem,consideringthedifferencebetweentraditionalmachinelearninganddfferentideasofensemblelearningal-gorithms,integratedlinearmodelandnonlinearmodel,linearregression,randomforestandGBDTwereinvolvedinbase-learnerlayer,andthemodelbasedonmulti-algorithmcombinationofStackingensemblelearningwasbuilt,accuracyandeffectivenessoftheproposedmethodwasconfirmedbasedonthedataoftransformerdistrictswithDG.Keywords:transformerdistrictwithDG,linelossrate,mutualinformation,ensembl...
