电力系统及其自动化学报ProceedingsoftheCSU-EPSA第35卷第4期2023年4月Vol.35No.4Apr.2023多端直流电网的非边界依赖型线路故障辨识方案周泽雷1,吴正平1,付翊航2,刘琦1(1.三峡大学电气与新能源学院,宜昌443002;2.国网湖北直流运检公司,宜昌443002)摘要:为实现在网孔结构多端直流电网中仅利用本地测量信息进行故障辨识,本文提出一种基于多频带能量分布特征和模式识别技术的故障辨识方案。首先,利用小波包将故障暂态量分解为多能量频带,然后构建深度置信网络模型来分辨不同频带能量分布之间的差异性,模型分辨出的故障数据组可对应不同的故障类型,结合具有方向识别功能的启动判据形成一套完整的非边界依赖型直流线路故障辨识方案。仿真结果表明,该方案能快速、有效地辨识区内、区外故障。关键词:网孔结构;多端直流电网;频带能量;深度置信网络;故障辨识中图分类号:TM75文献标志码:A文章编号:1003-8930(2023)04-0100-09DOI:10.19635/j.cnki.csu-epsa.001202LineFaultIdentificationSchemeforMulti-terminalDCGridIndependentofBoundaryComponentZHOUZelei1,WUZhengping1,FUYihang2,LIUQi1(1.CollegeofElectricalEngineering&NewEnergy,ChinaThreeGorgesUniversity,Yichang443002,China;2.StateGridHubeiDCTransportationInspectionCompany,Yichang443002,China)Abstract:Torealizefaultidentificationinamulti-terminalDCgridwithameshstructureusingonlythelocalmeasure⁃mentinformation,afaultidentificationschemebasedontheenergydistributioncharacteristicsinmultiplefrequencybandsandthepatternrecognitiontechnologyisproposedinthispaper.First,waveletpacketisusedtodecomposethefaulttransientintotheenergyofmultiplefrequencybands.Then,adeepbeliefnetworkmodelisbuilttodistinguishthedifferenceinenergydistributionbetweendifferentfrequencybands.Thefaultdatasetsidentifiedbythismodelcancor⁃respondtodifferentfaulttypes,andacompletesetofDClinefaultidentificationschemeindependentofboundarycom⁃ponentisformedbycombiningthestartingcriterionwithadirectionrecognitionfunction.Simulationresultsshowthattheproposedschemecanidentifythefaultsinandoutoftheareaquicklyandeffectively.Keywords:meshstructure;multi-terminalDCgrid;energyoffrequencyband;deepbeliefnetw...