文档名：基于物联感知数据和张量融合的电力变压器绕组绝缘劣化评估方法
摘要：针对目前缺乏依托在线物联感知数据的电力变压器绕组绝缘劣化评估方法的问题,该文考虑电、热、机械劣化因素对电力变压器绕组绝缘的损伤累积效应,提出基于物联感知数据和张量融合的电力变压器绕组绝缘劣化评估方法.首先,研究电、热、机械因素对绕组绝缘的累积损伤机理,依托电压、电流、温度、局部放电物联感知数据构建变压器绕组绝缘的电性能、热性能和机械性能劣化损伤指标;然后,构建三种劣化损伤指标的特征张量,基于张量融合对三种劣化损伤指标进行特征融合,提取劣化损伤指标间的高维劣化特征关联信息;最后,采用基于自组织映射网络(SOM)的最小量化误差方法构建绕组绝缘的综合劣化评估指标,实现对绕组绝缘劣化状态的评估.该文通过多种评价准则评价构建综合劣化指标,算例结果表明,所提方法能准确评估变压器绕组绝缘的真实劣化程度.
Abstract：Onlineevaluationofwindinginsulationdegradationisofgreatsignificancetothestableoperationoftransformers.Duetothelackofcorrespondingonlinesensingmeans,thetraditionaldegradationevaluationmethodsrelyingondatasuchasfurfuralcontentandmethanolcontentinoilcannotrealizetheonlineevaluationofwindinginsulationdegradation.Mostofthedegradationevaluationmethodsbasedononlinesensingdata,suchasvoltageandcurrent,onlyconsidertheinfluenceofasinglefactor.Itisdifficulttofullyreflectthedegradationdegreeofwindinginsulation.Therefore,thispaperproposesapowertransformerwindinginsulationdegradationevaluationmethodbasedonIoTsensingdataandtensorfusion,consideringtheinfluenceofelectrical,thermal,andmechanicalfactorsoninsulationdegradation.Itreliesonvoltage,current,temperature,andpartialdischargeIoTsensingdatatorealizeonlineevaluationoftransformerwindinginsulationdegradation.Firstly,thecumulativedamagemechanismofwindinginsulationdegradationcausedbyelectrical,thermal,andmechanicalfactorsisanalyzed.ThetransformerIoTsensingdataofvoltage,current,temperature,andpartialdischargeisusedtoconstructtheelectrical,thermal,andmechanicalperformancedegradationdamageindicatorsofwindinginsulation.Then,basedontensorfusion,featurefusionofthreedegradationdamageindicatorsisperformed,andhigh-dimensionaldegradationfeaturecorrelationinformationbetweendegradationdamageindicatorsisextracted.Finally,theminimumquantizationerrorofaself-organizingmapisusedtoquantifythedistancebetweenthedegradationfeatureoutputtensorandthebestmatchingunitweighttensor.Acomprehensivedegradationevaluationindexisthenconstructed,andtheonlineevaluationofthewindinginsulationdegradationdegreeisrealized.Basedontheacceleratedagingtestdata,thetrendevaluationvalue,monotonicityevaluationvalue,robustnessevaluationvalue,scalesimilarityevaluationvalue,andfusionevaluationvalueofthecomprehensivedegradationevaluationindexare95.78%,100.00%,99.75%,93.24%,and97.19%,respectively.Themeanabsolute,meansquare,androotmeansquareerrorsofthecomprehensivedegradationevaluationindexarelessthan0.05.TheR-SquareandPearsoncorrelationcoefficientsexceed97%,andthesignificancetestcoefficientislessthan0.01.Comparedwithtraditionaldegradationevaluationmethods,thefusionevaluationvalueoftheproposedmethodexceeds95%andhasahighersimilarityintrendwiththefurfuralcontentindexofwindinginsulation.Theconclusionsareasfollows:(1)TheproposedmethodreliesonthetransformerIoTsensingdatatoquantifythecumulativedamagecausedbyelectrical,thermal,andmechanicalstressesonwindinginsulation.Itcanaccuratelydescribethedegradationtrendofwindinginsulationundermultiplestresses.(2)Theproposedmethodfusesthefeaturetensorsofelectrical,thermal,andmechanicaldegradationdamageindicatorsbasedontensorfusion,whichcanextractthehigh-dimensionaldegradationcorrelationinformationbetweendegradationdamageindicatorstothegreatestextentwhileretainingthecharacteristicsofeachdegradationdamage.(3)Theproposedmethodcanaccuratelydescribethedegreeofdeviationbetweenthewindinginsulationdegradationstateandthehealthystatethroughthecomprehensivedegradationevaluationindexconstructedbytheminimumquantizationerror.(4)Accordingtotheexperimentalresults,theproposedmethodcanaccuratelyevaluatetheactualdegradationstateofthewindinginsulationbasedonthetransformerIoTsensingdata,andtheevaluationresultscanprovideareferenceforthemaintenanceofthewindinginsulation.
作者：曲岳晗 赵洪山 程晶煜 马利波 米增强Author：QuYuehan ZhaoHongshan ChengJingyu MaLibo MiZengqiang
作者单位：华北电力大学电气与电子工程学院保定071003
刊名：电工技术学报
Journal：TransactionsofChinaElectrotechnicalSociety
年，卷(期)：2024, 39(4)
分类号：TM41
关键词：电力变压器 绕组绝缘 劣化评估 物联感知数据 张量融合
Keywords：Powertransformer windinginsulation degradationevaluation IoTsensingdata tensorfusion
机标分类号：TU393.3TP391.41U442.55
在线出版日期：2024年3月5日
基金项目：国家国际科技合作专项基金基于物联感知数据和张量融合的电力变压器绕组绝缘劣化评估方法[
期刊论文] 电工技术学报--2024, 39(4)曲岳晗 赵洪山 程晶煜 马利波 米增强针对目前缺乏依托在线物联感知数据的电力变压器绕组绝缘劣化评估方法的问题,该文考虑电、热、机械劣化因素对电力变压器绕组绝缘的损伤累积效应,提出基于物联感知数据和张量融合的电力变压器绕组绝缘劣化评估方法.首先,研...参考文献和引证文献
参考文献
引证文献
本文读者也读过
相似文献
相关博文
基于物联感知数据和张量融合的电力变压器绕组绝缘劣化评估方法 Evaluation Method for Power Transformer Winding Insulation Degradation Based on IoT Sensing Data and Tensor Fusion
基于物联感知数据和张量融合的电力变压器绕组绝缘劣化评估方法.pdf
- 文件大小:
- 1.07 MB
- 下载次数:
- 60
-
高速下载
|
|
