基于支持向量机增量算法的锅炉燃烧效率建模研究

44220133BOILERTECHNOLOGYV0144NO2Mar20l3122(15106002430072)(SVM)1000MwSVMTK2296A16724763(2013)020001040(SVM)[1][21[331000MWSVM1n]11Karush-kuhnTucker[53a()zKarush-kuhnTucker(KKT)[6]fCt0yf(z)1OaCyf(z)1(1)lCyf(x)1Lagrange0oaCaC(1)0l()I1oaCl(a7)l1(2)liCIf(x)l1KKT120121017(61170024)(121031)(1963)44112SVMKKTKKTXnXVX5vXKKTXKKTXrKKT8vVXVKKTx?XPXXnSV222SVM21SVM3^1__SVM]F__7-tI3SVMmin()||W02Y(z)z+6ytWXt6i--1,2,-z(3)WXi+6Y6yi((z)18j1lmin()J|W+c(8+(4ul=1fYWX--be+WTi+b--yie+8e0i12ZC0(5)CeelImax(OlI)(6)ee4400e4i7L(68)1||W||z+c(l8+)-2=l2a(e+8Y+z+6)fi(++yizi--b)l(+'7ii)(7)l=1Wb8{--ZU--kOLlCa0azOLCi0i(8)(7)maxW(a)(i)()(z)+y()(+i)(9)K(xz)maxW({2)=1(ai)(--(2j)K(xz)+Y(a--{2)e(+i)(10)(z)(i)K(zz)+b(11)22SVMKK(exp()(12)Ce31000MW356(AAEF)624DCS30S31SVM19910120010000001000001C1000001CCrrili1-050100150200250300C5c5C(MAPE)5MAPECC225MAPE1100090O85080=075070065060|{h|0051515252535o6aO987654321444C22566194194MAPE194CC232321107043486322ASMEdASMEASME145200806660166SVM+^Xli}7ZnV20406080100120140160180200841000MWSVMSVM5l[1]dx--[J]201030(4)149152[2][J]200828(11)8286[3][c]2008[4][D]2011[5]WangLipoLiuBingWanChunruClassificationusingsupportvectormachineswithgradedresolution[J]GranularComputing20052(7)666670(16)21644DiscriminateandDiscussonsomeMistakeRegionandDoubtfulPointWheretheGasAcidDewpointCalculatingMethodshadbeStudiesZHANGJianzhong(ChinaPowerConstructionEngineeringConsultingCorporationBeijing100120China)AbstractThemistakeregionwherequotedandusedindomesticdocumentsonthegasaciddewpointtemperature(ADT)calculatingmethodsfromabroadhadbediscriminatedthedoubtfulpointofsomecalculatingmethodsfromabroaddocumentsofADThadbesurvedandanalyzedthemodellinecharacteristicpatternofvariousrelationcurveforADTcalculatingmethodscurrentlyandthescopeofapplicationincomparehadbenewstudiedKeywordsaciddewpointcalculatingS03contentmistakeregiondoubtfulpointclarificationdewpointcurvemodellinecharacteristic(4)1-63TRASVM[J]201031(3)609679[7]FengJunkaiYueGuangxiNewachievementsofdevelopResearchontheModelingofBomentofCFBCboilersinChina[C]ProceedingsoftheinternationalofconferenceonenergyandenvironmentShanghai1998505510ilerCombustionEfficiencyBasedontheIncrementalAlgorithmofSupportVectorMachineYEXiang-qianlTANLei2FANGYanjun2(1GuangdongPowerGridCorporationGuangzhou510600China2WuhanUniversitytheSchoolofthePowerandMechanicalWuhan430072China)AbstractTosolvetheproblemofboilercombustionefficiencymodelingvariousparameterssuchasfuelflowandairflowareselectedtodescribetheimpactofloadontheboilerthermalefficiencyandthetwostagecombustionmodelfortheboilerareestablishedbasedOnsupportvectormachine(SVM)ThenthecarboncontentofflyashandboilercombustionefficiencyarepredictedaccordingtothehistoricalconditionstherebytheimprovedincrementallearningalgorithmcanbeutilizedtooptimizethetrainingsampleswhichcanensurethepredictionaccuracyandreducethetrainingtimeThefieldoperatingdataof1000MWunitofSanbaimenPowerPlantinthecityofChaozhouDatangofGuangdongprovinceisutilizedtoestablishtheSVMmodelforthecarboncontentofflyashandboilerthermalefficiencyThesimulationanalysisindicatesthatthereiS1ittledifferencebetweenthepredictedvalueandmeasuredvalueofthemodeIwhichshowsthiscombustionefficiencymodelhasgoodpredictiveeffectKeywordscombustionefficiencycombustionmodelsupportvectormachine