外文翻译---高速铣加工中轴转矩的模糊控制.doc

附录A：英文资料TechnicalBriefsFuzzyControlofSpindleTorqueinHigh-SpeedMillingProcessesRodolfoHaber-GuerraInstitutodeAutomáticaIndustrial(CSIC),kin.22800N-III,LaProveda,28500Madrid,SpainStevenY.LiangTheGeorgeW.WoodruffSchoolofMechanicalEngineering,GeorgiaInstituteofTechnology,801FerstDrive,N.W.,Atlanta,GA30332-0405JoseR.AliqueInstitutodeAutomáticaIndustrial(CSIC),km.22800N-III,LaProveda,28500Madrid,SpainRodolfoHaber-HaberUniversidaddeOriente,Ave.Americass/n.,SantiagodeCuba,90400CubaThispaperpresentsthedesignandimplementationofatwo-input/two-outputfuzzylogic-basedtorquecontrolsystemembeddedinanopenarchitecturecomputernumericalcontrol(CNC)foroptimizingthematerialremovalrateinhigh-speedmillingprocesses.ThecontrolsystemadjuststhefeedrateandspindlespeedsimultaneouslyasneededtoregulatethecuttingtorqueusingtheCNCsownresources.Thecontrolsystemconsistsofatwo-input(i.e.,torqueerrorandchangeoferror),two-output(i.e.,feedrateandspindlespeedincrement)fuzzycontroller；whichisembeddedwithinkernelofastandardopencontrol.Twoapproachesaretested,andtheirperformanceisassessedusingseveralperformancemeasurements.Theseapproachesareatwo-input/two-outputfuzzycontrollerandasingle-output(i.e.,feedratemodificationonly)fuzzycontroller.Theresultsdemonstratethattheproposedcontrolstrategyprovidesbetteraccuracyandmachiningcycletimethanotherstrategies,thusincreasingthemetalremovalrate.DOI:I0.1115/1.2194063Keywords:fuzzycontrol,torque,high-speedmillingContributedbytheManufacturingEngineeringDivisionofASMEforpublicationintheJOURNALOFMANUFACTURINGSCIENCEANDENGINEERING.ManuscriptreceivedAugust19,2005:finalrevisionreceivedFebruary14,2006.ReviewconductedbyC.J.Li.IntroductionInordertoimprovemachiningefficiencyinahigh-speedmillingprocessthroughahighermaterialremovalrate,thisstudyfocusesonthedesignandimplementationofatwo-input/two-output(TITO)fuzzycontrolsystemforspindletorque.Themajorissuetobedealtwithisthenewdevelopmentandapplicationoffuzzylogic(FL)usingtheCNCsownresources.Noadditionalhardwareoverheadisrequired,sincethecontrolalgorithmisembeddedwithinthekernelofastandardopencontrol.Fuzzylogic(FL)wasselectedoutofalltheavailabletechniques,becauseithasprovenusefulasahighlypracticaloptimizingtoolforcontrolandindustrialengineering.Tothebestofourknowledge,themainadvantageofthisapproachisthatitincludes:(i)afuzzycontrollerembeddedinanopenarchitectureCNCtodealwiththeproductionenvironment；(ii)asimplecomputationalprocedureforfulfillingthetimerequirements；and(iii)norestrictionsintermsofsensorcost(thetorquesignalisprovidedbytheopenCNC),wiring,orsynchronizationwiththeCNC.Thispaperisorganizedasfollows:InSec.wepresentabriefstudyofamechanisticmodelforpredictingcuttingforceandspindletorque；inSec.IIIwedescribethedesignofthefuzzycontrollertooptimizethemillingprocess；inSec.IVwedescribehowthefuzzycontrollercanbeembeddedinopenarchitectureCNCs,andwediscussthekeydesignandprogrammingstages；inSec.Vwereviewtheexperimentalresultsandexploresomeofthecomparativestudies.Finally,wepresentconclusionsinSec.VI.High-SpeedMilling-ProcessControlBasedontheSpindleTorqueSignalThemechanisticmodelestimatescutting-forcevectorsandspindletorqueonthebasisoffeedrate,spindlespeed,andmaterialconstants.Themechanisticmodelofendmillingimplementedinthisresearchisbasedon1,2.Inordertomakeasimplifiedmodel,weusedthefollowingapproximation3,4:Ifthefeed-per-toothvalueftjissmallwithregardtothetoolradiusD/2,thenforatoolfedinthepositiveXdirection,theinstantaneouschipthicknesshjisrelatedtotherotationangleofthecuttingtool.Asthecuttingtoolrotates,chipthicknessvariesasafunctionoftheradialangle()andaxialangle()ofthecuttingtoolhj(j)=fijsinj·sin(1)Therefore,onthebasisoftheaboveapproximation,andforroughingandsemifinishedcutting,itisnotnecessarytoextracttheinstantaneouschipthicknessfromthevolumeinformation.Certainly,acorrectvalueofthefeedpertoothfijnotonlyincreasestoolperformancebutalsoimprovestheefficiencyofthemachiningprocess.Foracylindricalendmill,thefollowingconditionsaredefinedforfindingthegeneralsolution.r(z)=D2=90º(2)=k0zk0=(2tan)/Dwhere(z)isthelaganglethatappearsduetothehelixangleofthecuttingtool.Thisangleisconstantinthecaseofacylindricalendmill,anditvariesforaball-endmill.Thetangential,radial,andaxialforcedifferentials(dFt),(dFr),(dFa)actonaninfinitesimallengthdSofthecuttingedgeofthetool4dFt=KtedS+Ktchj(,)dbdFr=KredS+Krchj(,)db(3)dFa=KaedS+Kachj(,)dbItisalsoconsideredthatdb=dzsin(4)Furthermore,thecharacteristicsofapointonthecuttingsurfaceareidentifiedusingthepropertiesofkinematicrigidityandthedisplacementsbetweenthetoolandtheworkpiece.Theconstantsorcuttingcoefficients(Ktc,Krc,Kac,Kte,Kre,Kae)canbefoundexperimentallyusingcuttingforcespertoothaveragedforaspecifictypeoftoolandmaterial5.ThetotalcuttingforcesasafunctionofalongtheaxialdepthofcutforallthecuttingedgesthatareincontactwiththeworkpiececanbecalculatedasNfNfz2Fx()=(Fxj(j(z)=-dFrjsinjsinj-dFijcosj-dFajsinjcosjdzj=1j=1z1NfNfz2(5)Fy()=(Fyj(j(z)=-dFrjcosjsinj-dFijsinj-dFajcosjcosjdzj=1j=1z1NfNfz2Fz()=(Fzj(j(z)=-dFrjcosj-dFajsinjdzj=1j=1z1wherez1andz2aretheintegrationlimitsandFx(),Fy(),Fz()aretheresultingforcesforeachaxis.CuttingtorqueTqeisestimatedonthebasisofthetangentialforcedifferential(dFt)andthetooldiameter(D).TheoverallcuttingtorqueisTqe=Ft·D/2(6)FuzzyControlofSpindleTorqueinaHigh-SpeedMillingProcessThemanipulated(action)variablesweselectedwerethefeedrateincrement(fasapercentageoftheinitialvalueprogrammedintotheCNC)andthespindlespeedincrement(sasapercentageoftheinitialvalueprogrammedintotheCNC).Thethreebasictasksknownasfuzzification,decisionmaking,anddefuzzificationwereused.TheerrorandoutputvectorswereeT=KE·TqKCE·2TqFig.1Fuzzypartitionsandmembershipfunctionsfor(a)Tq,2Tq,(b)f,and(c)su=GC·fs