您好,欢迎访问三七文档
当前位置:首页 > 行业资料 > 国内外标准规范 > 基于条件信息熵与贝叶斯网络的变压器故障诊断研究
,,,(200090:,。:;;;:TM406:B:1001-8425(2009)07-0048-04ResearchonTransformerFaultDiagnosisBasedonConditionalInformationEntropyandBayesianNetworkJIANGYu-rong,ZHUFan,FUYang,CAOJia-lin(ShanghaiUniversityofElectricPower,Shanghai200090,China)Abstract:Themethodoftransformerfaultdiagnosisbasedonconditionalinformationen-tropyandBayesiannetworkispresented.Thefaultexamplesareanalyzed.Keywords:Transformer;Faultdiagnosis;Conditionalinformationentropy;Bayesiannet-work1,。、,,。,。,,。22.1U,U(、)Uσ,。1:P、QUX,Y(X={X1,X2,…,Xn},Y={Y1,Y2,…,Yn}),P、QUσ:X:::p=X1X2…Xnp(X1)p(X2)…p(Xn::)Y:::p=Y1Y2…Ymp(Y1)p(Y2)…p(Ym::),p(Xi)=|Xi||U|,i=1,2,…,n;p(Yi)=|Yi||U|,i=1,2,…,m2:()PH(P):H(P)=-ni=1Σp(Xi)log[p(Xi)]。3:()Q[U|IND(Q)]={Y1,Y2,…,Ym}P[U|IND(P)]={X1,X2,…,Xm}():H(Q|P)=-ni=1Σp(Xi)mj=1Σp(Yi|Xi)log[p(Yi|Xi)]:p(Yi|Xi)=|Yi∩Xi|/|Xi|,i=1,2,…,n,j=1,2,…,m。:(061612040);(P1301);(08ZZ92)46720097TRANSFORMERVol.46JulyNo.72009、、7PD={d}U(|U|=n):U|IND(P)={X1,X2,…,Xt};U|IND(d)={Y1,Y2,…,Ym}1:U,PU,d,UP{d},PrPd(),H({d}|P)=H({d}|P-{r})。2.2,,,,,。CEBARKNC,:DaH(D|{a})a,H(D|{a}),a,,。:(1)DCH(D|C)。(2)H(D|{ai}(ai∈C),aiH(D|{ai})。(3)B=CH(D|{ai})ai:①BaiH(D|B-{ai})。②H(D|C)=H(D|B-{ai}),ai,B=B-{ai};,ai,B。3(、)。。:U={X1,X2,…,Xn},Xi。,B=〈G,Θ〉。G,X1,X2,…,Xn,。BU:P(X1,X2,…,Xn)=i仪P(Xi|Pai)1()。3:1,。2,。:P(X)=ni=1仪P(Xi|x1,x2,…,xi-1)PaiXi,P(X1,X2,…,Xn)=i仪P(Xi|Pai)3,P(Xi|Pai)。,Xi。4,,。4.1,M、D、Pi、Cij,12。2,3。31,0;1~10。(:):V1V2V3V4V5V61Fig.1Bayesiannetworkstructurediagram49462D、Cij、PiTable2ListoffaultsetD,causestrengthCijandpriorprobabilityPi/%d1m10.9022.71m20.818m50.189m70.30d2m20.2195.27m40.267m80.816d3m20.7136.29m50.289m70.35d4m50.5156.12m60.80m70.90m80.681d5m40.7185.06m90.75d6m20.67413.17m30.87m50.231d7m50.8637.98m70.90d8m40.41614.2m50.879m70.90m80.759d9m20.14912.02m50.681m60.80m70.75m80.72d10m20.207.17m40.60m50.70m70.901MTable1ListofsymptomsetMm1m2m3m4m5m6m7m8φ(CO)/φ(CO2)m9(1)H(D|M)。(2)H(D|{ai}),,4。(3)H(D|M-{mi}),C′={m1,m2,m4,m7,m8},(5),。4.25,2,D,m1、m2、m4、m7m8。3Table3Listoffaultdiagnosisdecisionm1m2m3m4m5m6m7m8m9did11100101001(22.71%)d20101000102(5.27%)d30100101003(6.29%)d40000111104(6.12%)d50001000015(5.06%)d60110100006(13.17%)d70000101007(7.98%)d80001101108(14.21%)d90100111109(12.02%)d1001011010010(7.17%)4Table4Taxisresultsm9m6m5m3m1m7m4m2m80.8590.8460.7950.7760.7130.7090.6740.6690.6585Table5Listofminimumdecisionoffaultdiagnosism1m2m4m7m8d111010d201101d301010d400011d500100d601000d700010d800111d901011d100111050、、77Table7Probabilitesoffaultcausesd1d2d3d4d5d6d7d8d9d10P0.02200.00060.0137000.0870000.00110.0003[6],,M+()M-(),M=M++M-。,。:(1):P(di|M+∧M-)=P(M+∧M-|di)·P(di)P(M+∧M-)(2),,max{P(di|M+∧M-)}。52SFP-120000/220,,DGA6。M+={m2},M-={m1,m4,m7,m8},P(di|M+∧M-)7。P(d6|M+∧M-),;:,,,,,。。6,,,,,,。,,,。:[1].[J].,2003,19(3):1-5.[2],,.[J].,2006,26(8):137-141.[3],,.[J].,2002,25(7):759-766.[4],,,.[J].,1999,33(4):13-16.[5],,,.[M].:,1993.[6],,.[J].,2005,20(4):45-51.d1d2d3d4d5d6d7d8d9d10m1m2m4m7m82Fig.2FaultdiagnosismodelwithBayesiannetwork6DGATable6DGAdataoffaulttransformerH2CH4C2H6C2H4C2H2C1+C2COCO21996-12-07144.71.45.611.737739601997-02-05133238.958476.76.757302422715:2008-07-08:(1976-),,,,;(1983-),,,,;(1968-),,,,、。51
本文标题:基于条件信息熵与贝叶斯网络的变压器故障诊断研究
链接地址:https://www.777doc.com/doc-9199606 .html