旋转机械故障诊断的灰度_基元共生矩阵方法研究

23920089JournalofAerospacePowerVol.23No.9Sept.2008:100028055(2008)09216092082,,(,150001):2,,,BP600MW,.:;;2;;:TK473;TB53:A:2007209215;:2007212220:(1977-),,,,.Faultdiagnosisofrotatingmachinerybasedongrayprimitiveco2occurrencematrixDOUWei,LIUZhan2sheng,MAXiao2feng(SchoolofEnergyScienceandEngineering,HarbinInstituteofTechnology,Harbin150001,China)Abstract:Afaultdiagnosismethodbasedongrayprimitiveco2occurrencematrixwasanalyzed.Thismethoddirectlyextractstheimagecharacteristicsaboutthevibrationparame2terofrotatingmachinerybycombiningrayspacedistributioncharacteristicswithtextureprimitivemethod.Whileextractingtheinformationofimagetexturecharacteristic,itena2blesfaultdiagnosisofrotatingmachinerybyusingBP(backpropagation)artificialneuralnetworks.Thismethodwasvalidatedtogethighdiagnosisaccuracybyconductingnormalstaterotor,unbalancedrotor,misalignmentrotorandloosebearingpedestalexperimentson600MWturbineexperimentalbench.Keywords:rotatingmachinery;faultdiagnosis;grayprimitiveco2occurrencematrix;parametersimage;texturefeature,,,[1].,,,,..,.,,,.[2].,223,,,,(),,.600MW,,.12,.,..,2.,,.1.12,h(i,j|d,),dij(),[3211].,h(i,j|d,).,.,.,48.2,.1,.,,().1.1.1f(i,j),i=0,1,2,,Lx-1;j=1,2,,Ly-11(a)1(b);.1Fig.1Schematicdiagramsofprimitiveframework,4(1(a))m(i,j)=f(i-1,j)+f(i,j+1)+f(i+1,j)+f(i,j-1)(i=1,2,,Lx-2;j=1,2,,Ly-2)(1)8(1(b))m(i,j)=INT{[f(i-1,j-1)+f(i-1,j+1)+f(i+1,j-1)+f(i+1,j+1)]2[f(i-1,j)+f(i,j+1)+f(i+1,j)+f(i,j-1)]}(i=1,2,,Lx-2;j=1,2,,Ly-2)(2)INT.1(a),2(a).k(1),m(i,j)=k=1k[f(i-k,j)+f(i+k,j)+f(i,j-k)+f(i,j+k)](3),2(b),(c),(d)..m(i,j),P={m(i,j)|i=1,2,,Lx-2;j=1,2,,Ly-2}.2Fig.2Schematicdiagramsoftextureprimitiveextended1.1.2.,,.fmax,fnormalize,01619:2F(i,j)=INTf(i,j)fnormalizefmax+1(4),0fmax0fnormalize,,.,mmax,mnormalize,P(i,j)=INTm(i,j)mnormalizemmax+1(5),2F(i,j)(i=1,2,,Lx-2;j=1,2,,Ly-2)P(i,j)(i=1,2,,Lx-2;j=1,2,,Ly-2)1.1.32fnormalize=Lgpx,mnormalize=Lgpy.2GP(i,j),i=0,1,2,,Lgpx;j=0,1,2,,Lgpy,[529].(Lgpx+1)(Lgpy+1).GP(i,j)i,j.GP(6,10)=18,2,6(i=6),10(j=10)18.3,490.(6)3(0),(7)4(0);(8)(9)344,(10)342.,;(90),2.,.(1(a),(b),(d)).,.2(10)(11);,,.h(i,j|1,0)=22402428130122223216(6)h(i,j|1,0)=186046146406128844812(7)p=004221664521889745999789777910115577911(8)p=579860579860779944797752910842211119511(9)GP=222010200000001014020000000000041410000010012302(10)GP=GP(11)1.22,,,6.6:(E),116123(H),(L),(I),(UV).:E=ij[GP(i,j)]2(12)H=-ijGP(i,j)logGP(i,j)(13)L=ij11+(i-j)2GP(i,j)(14)I=ij(i-j)2GP(i,j)(15)U=ijGP(i,j)2ijGP(i,j)(16)V=ijGP(i,j)T2ijGP(i,j)(17)6t=[E,H,L,I,U,V](18)6,,,,.,,,[12].[t1,t2,,t6],[T1,T2,,T6],:m.Ti=(ti-m)/,i=1,2,,5(19)2,BP,5.5BPFig.5ModelofBPneuralnetworksBPnm[13],,,,BP.Sigmoid.BP,BP,,BP,.BP,,,,,,,2,,.,BP,.,.6.,,.,.3,600MW5.7,600MW5.8600MW.600MW95(7).55KWFRENIC,HG0G2C2.,21619:26Fig.6BlockdiagramoffaultdiagnosismethodbasedonneuralnetworksBENTLY3000XL8mm,7187V/mm,,32,0.64s,3200r/min.A/D.,540,200,20,20.,,(),,9135,,,,,,,,.,,1415.,,2.27600MWFig.7Buildupofrotor2bearingsystemtest2bedof600MWsupercriticalsteamunitturbine31612341619:215Fig.15Grayimageofbearingpedestalloosenessvibration,6.BP.BP,,,6,6.5,,[1,0,0,0,0];[0,1,0,0,0],[0,0,1,0,0],[0,0,0,1,0][0,0,0,0,1],6,5,,:n1=(n+m)1/2+l(20)n1,m,n,l110.8,0.7,0.1,0.001,10000,62825BP.20,,,2024433,90.0%,80.0%,80.0%,85.0%85%,80%,,.2.3,2,,.2,BP.,.:[1],,.[J].,2001,20(4):36241.FENGZhipeng,SONGXigeng,XUEDongxin.Surveyofvi2brationfaultdiagnosisofrotationalmachinery[J].JournalofVibrationandShock,2001,20(4):36241.(inChinese)[2],.[J].,2007,33(14):1712173.ZHANGHengbo,OUZongying.Imageretrievalmethodbasedoncolorprimitiveco2occurrencematrix[J].ComputerEngi2neering,2007,33(14):1712173.(inChinese)[3],,,.[J].,2004,23(10):27229.ZHAOHui,BAOLi,LIANGGuangming,etal.Researchonmicro2cellimagebasedonsynthesizedgraylevelco2occurrencematrix[J].PatternRecognitionandSimulation,2004,23(10):27229.(inChinese)[4],,.[J].,2006,38(9):141921421.ZHANGLamei,WANGGuoxi,ZHUANGXuejing.Texturefeatheranalysisofthemembranewrinkling[J].JournalofHarbinInstituteofTechnology,2006,38(9):141921421.(inChinese)[5]JobanputraR,ClausiDA.Preservingboundariesforimagetexturesegmentationusinggreylevelco2occurringprobabilities[J].PatternRecognition,2006,39(2):2342245.[6]SuralVS,MajumdarAK.Anintegratedcolorandintensityco2occurrencematrix[J].PatternRecognitionLetters,2007,28(8):9742983.[7],.[J].,2004,(8):11214.ZHANGYunbin,ZHANGYongsheng.Anefficientalgorithmforimageretrievalbasedontexturefeatures[J].HighTech2nologyLetters,2004,(8):11214.(inChinese)[8],,,.SVM[J].,2003,43(4):4752478.WANGLiangshen,OUZongying,SUTieming,etal.Con2tent2basedimageretrievalindatabaseusingSVMandgrayprimitiveco2occurrencematrix[J].JournalofDalianUniversityofTechnology,2003,43(4):4752478.(inChinese)[9]WANGLiangshen,OUZongying.Imagetextureanalysisbygray2primitivecooccurrencematrix[J].ComputerEngi2516123neer,2004,30(23):19221.[10].2[J].,1984,10(1):22225.HONGJiguang.Graylevel2gradientco2occurrencematrixtextureanalysismethod[J].ActaAutomaticaSinica,1984,10(1):22225.(inChinese)[11],,.2[J].,2002,18(1):21224.WANGLiangshen,WANGWenyou,WANGHongmei.Retrievingimagebasedongrey2primitivecooccurrencema2trix[J].JournalofLi