资源描述:
《基于细菌觅食算法优化电力变压器故障诊断技术》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、基于细菌觅食算法优化的电力变压器故障诊断技术董方旭1,咸日常1,咸日明2,李文强3,马雪锋3(1.山东理工大学电气与电子工程学院,山东淄博255000;2.山东汇能电气有限公司,山东淄博255000;3.山东省计量科学研究院,山东济南255014)摘要:针对支持向量机(SVM)分类性能受参数影响,且最优参数难以获取这一问题,提出一种基于细菌觅食算法(BFA)的电力变压器故障诊断模型的参数寻优方法。该方法以电力变压器油中特征气体含量作为状态评价样本,通过BFA寻找全局最优SVM参数解,构建k-折平均分类准确率目标函数,建立
2、变压器故障诊断模型。仿真结果表明,BFA对SVM最优参数的选取较遗传算法(GA)、粒子群算法(PSO)更迅速,且优化后的SVM电力变压器故障诊断模型具有更高的精确度;利用BFA优化方法建立的SVM电力变压器状态诊断模型,对IEC三比值法中无法判断的数据也可进行精确诊断。最后,通过实例分析,验证了方法的有效性。关键词:细菌觅食算法;支持向量机;参数优化;电力变压器;油中色谱分析;故障诊断中图分类号:TM406文献标识码:B文章编号:1001-1390(2018)00-0000-00Faultdiagnosistechnol
3、ogyofpowertransformerbasedonbacterialforagingalgorithmoptimizationDongFangxu1,XianRichang1,XianRiming2,LiWenqiang3,MaXuefeng3(1.SchoolofElectricalandElectronicEngineering,ShandongUniversityofTechnology,Zibo255000,Shandong,China.2.ShandongHuinengElectricCo.,Ltd.,.
4、,Zibo255000,Shandong255000,China.3.ShandongInstituteofMetrology,Jinan255014,China)Abstract::AimingattheproblemsthattheclassificationperformanceofSVMisaffectedbyparametersandtheoptimalparametersaredifficulttoobtain,aparameteroptimizationmethodbasedonbacterialforag
5、ingalgorithm(BFA)forfaultdiagnosismodelofpowertransformerfaultdiagnosismodelisproposedinthispaper.Thegascontentsoftransformeroilwerecollectedasevaluationsamples,thebestglobaloptimalsolutionofSVMwassearchedbyBFA,theobjectivefunctionofk-foldaverageclassificationacc
6、uracyratewasconstructed,andthefaultanalysisdiagnosismodelofoptimalSVMpowertransformerwereestablished.ThesimulationresultsshowthattheselectionofSVMoptimalparametersbyBFAismorerapidthanthatbyusinggeneticalgorithm(GA)andparticleswarmoptimization(PSO),andtheoptimized
7、SVMpowertransformerfaultdiagnosismodelhashigheraccuracy.TheSVMstatediagnosismodelofpowertransformerwhichisestablishedbyBFAoptimizationmethodcanaccuratelydiagnosethedatathatcannotbediagnosedbyIECthreeratiomethodmethods.Finally,theeffectivenessoftheproposedmethodis
8、verifiedbyanexampleanalysis.Keywords::bacterialforagingalgorithm,supportvectormachine,parameteroptimization,powertransformerdissolved,gasanalysis,faultdiagnosi