资源描述:
《2006 neural-network based analysis and prediction of a compressor characteristic performance map》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、APPLIEDENERGYAppliedEnergy84(2007)48–55www.elsevier.com/locate/apenergyNeural-networkbasedanalysisandpredictionofacompressor’scharacteristicperformancemapaa,*abYouhongYu,LingenChen,FengruiSun,ChihWuaPostgraduateSchool,NavalUniversityofEngineering,Wuhan430033,PRChinabMechanicalEngi
2、neeringDepartment,USNavalAcademy,Annapolis,MD21402,USAAvailableonline15June2006AbstractThedifficulties,duetoalackofinformationaboutstage-by-stageaxial-compressorperformance,areanalyzed.Toovercometheseissues,athree-layerback-propagationneural-networkappliedLevenberg–Marquardtalgorith
3、mispresentedanddiscussed.Theexperimentaldataprovidedbymanufacturersareusedfortheneural-networktraining.Throughtwicetraining,thecompressor’sperformancemapcanbepredicted.Theresultscanbeusedforthedevelopmentofanoff-designmodeloroveralldynamicsimulationofthebehaviourofagas-turbinepower
4、-plant.Ó2006ElsevierLtd.Allrightsreserved.Keywords:Compressor;Characteristicmap;Neural-network;Performanceprediction1.IntroductionTheincreasingsuccessofgas-turbinepower-plantsinindustrialandmarineapplicationsisowedpartlytotheirquickresponsestoloadvariations.Gas-turbinepower-plantp
5、erfor-manceunderISOconditions(burningareferencefuel,suchasnaturalgas,at150°C,atmo-sphericpressure,and60%relativehumidity)isinformationprovidedbymachinemanufactures.Nevertheless,gas-turbinepower-plantfrequentlyoperatesunderoff-designconditions,forinstance,atpartloadorunderdifferentat
6、mosphericconditions,andintheseconditions,thechangesofgas-turbinepower-plantperformancecanbedramatic.Inordertodescribetheoff-designperformanceofgas-turbinepower-plantaccurately,goodpredictions*Correspondingauthor.Tel.:+862783615046;fax:+862783638709.E-mailaddresses:lingenchen@hotmai
7、l.com,lgchenna@yahoo.com(L.Chen).0306-2619/$-seefrontmatterÓ2006ElsevierLtd.Allrightsreserved.doi:10.1016/j.apenergy.2006.04.005Y.Yuetal./AppliedEnergy84(2007)48–5549ofcharacteristiccurvesofgas-turbinepower-plantcomponentsareessential[1].Amajorproblem,however,isalackofinformationa
8、boutstage-by-stageperformance.The