摘要
Theknowledgeofflowregimesisveryimportantinthestudyofatwo-phaseflowsystem.AnewflowregimeidentificationmethodbasedonaProbabilityDensityFunction(PDF)andaneuralnetworkisproposedinthispaper.Theinstantaneousdifferentialpressuresignalsofahorizontalflowwereacquiredwithadifferentialpressuresensor.ThecharactersofdifferentialpressuresignalsfordifferentflowregimesareanalyzedwiththePDF.Then,fourcharacteristicparametersofthePDFcurvesaredefined,thepeaknumber(K1),themaximumpeakvalue(K2),thepeakposition(K3)andthePDFvariance(K4).Thecharacteristicvectorswhichconsistofthefourcharacteristicparametersastheinputvectorstraintheneuralnetworktoclassifytheflowregimes.Experimentalresultsshowthatthisnovelmethodforidentifyingair-watertwo-phaseflowregimeshastheadvantageswithahighaccuracyandafastresponse.Theresultsclearlydemonstratethatthisnewmethodcouldprovideanaccurateidentificationofflowregimes.
出版日期
2005年01月11日(中国期刊网平台首次上网日期,不代表论文的发表时间)