外文翻譯--基于案例和知識(shí)的注塑模具分型面設(shè)計(jì)技術(shù) 英文版.pdf

CHINESEJOURNALOFMECHANICALENGINEERFGVol.18,No.4,2005.609.YuTOngminLiGUanhUaDESIGNTECHNOLOGYFORINJECT10NMOLDRARTINGSURFACEBASEDONCASESANDKNOWLEDGELIYbUminAbstract:onthebasisoftheeomPrehensiveanalysisabouttheautomatiegenerationoftheinjeet-LanJianKeyLaratoryforPdsionandNon-traditionalMaehiningfehnologyofMiniatryofEducaUon,D+nUniVe=tyofTeChnologDalian116024,Chinalonmo,dpartmaJce,thePartingsurfacemathod.whihaduce5aoWledcsebasedreasonmg(CBR)mtothecomPter-alaeadeslgn15aescnDeooyeomolnmgwitntncacua,eharaeteristieininjeetionmolddesign,andthedesignProeessofeaseBbasedreasoningmethod15alsogiven.AeaselibraryIneludingtheIntormatlonorPartlngsurjaeeB5ouBrwBtntnemaexormainshaPefeates.Theautomatiedesignofthemo1dPartingsurfaceisrealizedeombinedwiththeforardBreasoningmethodandthe517larity50utionproeedure.Theruleknowlelelibrary15alsoimProved.Keywords:InjeetionmoldPaIntingsuri(cedesignCaseBbasedreasoning(CBR)Similtysolution0INTRODUCTIONBeingoneoftheimPortantstePsininjectiontheehoieeofPartingsurfaeehasandireetlyefeetmolddesign,onthegeneralframedesi.moldmakinZeost,thepinsurfaeemaehiningandthequalltyottnetlnalPlasncPartsmBmcbotneratlonalehoosingofthePartingsurfaee15thePreeonditionofasueeessfulinjeetionmolddesignandthebasisofrealizinginjeetionmoldingPartantomatieandsoPhistieateddesign.Inreeentyears,researehworkonthePartingsurfaeeandmoldeomPonentsautomatieallygeneration15oneofthehotsPotsanddifieultiesintheinjeetionmoldingCADfield,andmanyreseafchersaroundtheworldhavedoneremarkableworksabout1FRAMEVVORKOFKNOWLEDGEANDCASE-BASEDREASONINGPARTINGSURFACEAUTOMATICDESIGNthisissue.ButthosePriorworkstendtoeshaPeanditsPositionrelationsofPrtingsXPrCSurfacSSeonstrUetingsomebumericalmodelsbasedonthegeometryallBaroundbytheessentialanalysisofthePlastiePartn5shapestruetureeomPlexityofPlasticPkindsofshortcomingsinPracticec2B5.Duetothediversityandts.thosemethodsh8VesomeWithmuehexPerieneeandlitletheoryformoldingdesign,thePartingsurfacedesignmethodWhiehintrodueesknowledgeandeaseBbasedreasoning(CBR)intotheeomPuterBaideddesign15PresentedbyeombiningwiththedesignProeessofinjeetionmoldingdesign,50thatthemoldingPartseanbegeneratedautomatieally.KnowledgeandcaseBbasedreasoningmeansPlying517IarityjudglnentProeedureandanalogismtohandleanewdesigntaskbyeonstruetingknowledgelibraryandeaselibraryandresPectivereasoningruleonthebasisofeombtheknowledgedruleofPartingsurfacedesiSueeessfulenglneersPrineiPlesueeessfulaPproachdesizsoPdthsucce,)oxPenencTorgan1CallymB,tnlsmenewQeSlgnnotonlydofTfitsofmoldPartingsurfaeebutalsoineludestheexPerieneeofthosefonnereasesAsaresult,this15moreeonvenient,quiekandPraetiealefeetivecomParedtoanalysisandsolvingthePartingsurfacedesignbyconstrUetingnumeriealmodel.FUrnthermorethedeseriPtionofPartingsurfaeedesignexPerieneesareeasilyexPressedbythecasesratherthantheabstraetdesignrulesandexPressions.50theknowledgeandeaseBbasedreasoningPartingsurfaeedesignmethod15verysuitableformoldengineern5thoughtfashionand118lner.ReeeivedMay26,2004；receivedinrevisedformAugUst17,2005；aeeePtedAugust30,2005ThegeneralideaofknowledgeandcaseBbasedreasoningPaltingsurfaeedesign15tofoundaeaselibraryusingsuceessfulPartlngsurfaCedesignsintermofmoldedPartsntyPesandgeometriecharaeter.ThemainshaPefealiresofmoldedPareindexesofthislibrary.AftersearchingthesimilareasewiththePart,smainshaPefeaturefromtheeaselibrary,thePartingsurieecanbegeneratedbyjudgingfromthesimilarityprineiPleandthegivenreasoningstrategy.Atthesametime,theeorrelativeknowleeandPrineiPlesofPartingsurfaeedesignandthedesignern5exPerieneeseanbesunnnarizedintoknowledgelibraJ,whieheanbequeriedandconsultedbydesignersastheassistantinformationofPartingsueedesign.TheeaseeomPonentofthemethod15usedforthereasoningdesign,whiletheknowledgeonefortheguidaneeofdesignern5rationalseleetionofPartingsurice,TheknowledgeandcaseBbasedreasoningPartingsurfaeedesimainlyeonsistsofthefollowing:Construetingknowl-edgelibr/.Ai7ngatthePracticeofinjeetionmoldingdesign,aknowledgelibraryofPartingsuri(eedesignrules15foundedthrougheoneludingandintegratingtherelatedowledgeandrulesofPartingsUrfaeedesignandthedesignersnexPerienees.Construetingeaselibraryofpartingsurfaceposition.AeaselibraryofParntingsurfaeePosition15eonstruetedbasedonthetyPesofmoldedPartsandthePrevioussuceess7designeases,whieheonsistsofthedeseriPtionofeasesandthestruetUreofcaselibrary,ThekeyofdeseribingthePositionofPingsurface15tosortandorganizetheshaPefeaturesofcasemodelsanddeternunetheeoresPondlngPositonofPngsurfaceaceordingtothevariantsheofmoldedParts.BeeausethePartingsurfaeePosition15mainlydetenninedbythePart,5shaPefealre,theorganizationofeaselibraryshouldbeProPitioustotheeasesearchingindesinroeess.xdentifieationofmoldedpartnsshaPefeature.TheshaPefeatUresoforiginalPartshouldbeidentifiedbeforedetenIningthePositionofPartingsurfaee.Therefore,themainshaPefeatesindexneedtobefoundedaimingatthetyPeofmoldedPartstruetureThustheseatnlessjointbetweenmodelingmeldedPartanddesignmoduleofmoldPBngfaceoanberealized.Asaresult,theheritedcharacteristieofsystem15alsoimProved.Systemreasoningandsearehingmodel.SearehingandmatehingthesimilarshaPe.610.ttttitechnoloforinjeetionmoldpartsurfacebasedonCaseSandknowledgefoatureofPartmodelfromthePartingsurfaeePositioneases15thePrimarytaskofthePartingsurfaeePositionease,5searehingRecommendedstyleUnadyBsablestyleeourse.HeneebuildingtheshaPefeatureindexodmodel15eoretorealizeexaeteaseAutomatiegenerationreasoningandquieksearchtheingaofPartingsurfaee.AfterPerformingdesignmarehingandadjustingabouttheseleetedPartingsurfaeePositioneasewhieh15517lartotheshaPefeatureofmoldedPart,theneededmoldPartingsurfaeeeanbegeneratedautomatieally.storageofeases.Thoabunnteaselibr+ofPartingsurfaeePosition15theimPortantfoundationtoimprovethedesignqualityofmoldPartingsurfaceandreasoningefieieney.ThesystemProvidesstoragejudgmentandseleetionbasedonelassifieationbyeasefeaturesaboutthefinallyresultofeaehnewdesigntask.TheflowehartofknowledgeandeaseBbasedreasoningPingsurfaeedesigning15showninFig.1.Fig,2DemonstitionoftheknowledgelibraryofPartingsurfaeedesignModelpartfesFeatureldendcationMaFhingmF+sindexedshapemdthecasemodelCaselibbf51jlarsimilartothepart,eportsbe1po.lmBonDe1runethePingsurceusingsimil0ty501bbpedebaontheeghgbdeldtothecmlibmryofpg,WcepositonThePartingsurfaeedesignmainlyhasrelationtothemainshaPefeatureofmoldedPart,andtheidentifieationofthemainshefeatUrealsoPlaysaleadjngroleneasesearching,Ijsregard,thestrUetUreofeaselibraryshouldbeorganizedaeeordingtotheshaPefeatureofPart,andthehierarehiealfeaturetreeofmoldedPartshouldbeeonstrueted,bywhiehtheparteaseseanbeelassifiedstored.ClassifyingtheshaPefeatUresofPartsbytheirdiferenteharaeters,theeommonfaetorsofPartmodeleanbeextraetedastheParentnode.TheehildnodeeaninheritallofatributesFmtheParentnode,andeontainmorefeaturestliantheParentone.Asfarasasingleease15eoneerned,alongwiththeinheritedrelationgoingthroughlayerbylayer,thenumberoffeaeatributeswillinereaseThefeatureatrlbutesaremoreaPttodeseribingeoneretelyandbyinstanee,andtheeomPlexdegreeoftheirrelations15alsoinereased.ThiselassifiedProaeheanhelPtosearehtheeaselibrBindexedbyshaPefeaFresandtosolvethemodedPwitheomPlexshaPe.33.1CASEDESCRIPTIONSANDSEARCHINGCasedeseriPtions2BUILDINGOFKNOWLEDGELIBRARYANDCASELIBRARYTheknowledgelibrary15theilnPortantresourcethateanassistdesignertoseleetrationalPartingsurfaeePosition.TheorganizationanddeseriPtionofowledgeinthelibraryshouldfitforthePraetiealmolddesignProeessandeanbeeasilyqueriedandeonsultedbydesigners.Therefore,theeommonusedknowledgeandPrineiPlesinPartingsurfacedesignandthedesiguersnsueeessfulexPerieneeswereeoFludedandintegrated.SomePieeesofdesignrulesknowledge,whiehformtheknowledgelibraryofPartingsurfaeedesign,werestablishedandwereidentifiedbysimPlesentenees.AllofthedesignrulesinlibraryweredeseribedandexPressedindetailbyfigureandleter.Andtheelue15alsogiveninoreeoln1endedn,andonotrecommended,nwaysinthefigureeomPonentofdesignrules,whieheanhelPthedesignmaketherationalselecjonbasedonPartieulardesignease.Fig,2showsthePartingsurfaeePositiondeseriPtiongivenbythesystemafterehoosingoneofthedesignrulescTheeasedeseriPtion15thefoundationtoPerformaknowledgebasedreasoning,whiehtumstodeseribeanexistedsuceessfuldesigneaseasthedatainformationthateanbeidentifiedandProcessedbyeomPuteTheeontentandthedeseriPtivewayshaverelationwiththequalityandefieieneyofsolving517Iarquestions.TheeasedeseriPtionmaybeexPressedinthefollowingmode,easeofthemoldedPart=theindexinformation+PositioninfonationofPartingsurfaee,wheretheindexinformation15theshaPefeatureofthePart.NamelytheshaPefeatoreofPartasanindexeansearchthesimilareasewiththemainshaPefeature,andtheresPeetiveParntingsurfaeePosition15alsogiveninthisease.ThustheeasesinPartingsurfaeePositioneaselibraryaresometyPiealPartshaPeandtheeoresPondingPartingsurfaeePositiononthePartmodel.ThenumeriealdeseriPtion15ease(F3=ease(,.,l,(sJ,l,m)whereF=,.,=,),115theehildshaPefeatureintheinjeetionmoldingease,i=1,2,l,n,n15thenumberofehildfeaturesintheease,S=(sB,52,l,sb),sj15thesolvingscheme,i.e.thePartingsurfaeePosition.j=1,2,l,m,|15thenumberofavailablePartingsuri(eepositionofoneeasemodelcAhierarchiealeaselibra,whieh15showninFig3,eanbeformedaeeordingtothenumberoftheshaPefeaturesineases.AndthenextayereontalnsonemoreshaPefeatUrethanthefonerone.ItdeveloPstheshaPeinformationofeasemodelonthebasisofinheritingtheformerlayern5shaPefeatures.Fig4showstheorganizingsehemeofeomPlexeases.Theseheme15eonsisteniwiththemodelProeessofmoldedPart,thusit15convenienttoPerformingeasesearehingindexedbytheshefeatijre.3.2HierarchicalsearehingbasedontheProduetruleCasesehing15animPortanteomPonentinreasoningmeehanism.Theesseniialofsearehing15theProeessofsearchingandmatohing.50theProeessofeonstrueinginfereneemeehan-CHINESEJOURNALOFMECHANICALENGINEERING.611.ism15alsotheProcessofeonstIuetingthesearchingstrategiesandmatchingthesi7lties.Thestrategieshaveagteatefeetonthereasoningeficieneyandtheresult.ThePuJJoseofthesearchingofthePartingsurfacePositioncaseistosearchthecasewhichhasthesimilarmainshapefeaturewiththePlastiemodel.Thusthemethod15ProPosedonthebasisofProduetruleaeeordingtotheoanjzationalstrUetUreofeaselibrary,Thehierarchiealsearehingmeansthatthecaselibrary15searchedintu+fromtheheadlayertotheendinthesearehingProeedure,butwhiehembranchmentwouldbefollowedshouldbejudgedfromtheProduetrule.Fig.3sclireofthePartinogrganlzationlmsurfaeePositioneaseibrarymPI/atedcalsFig.4organizationstructureoftheeomPlicatedeaseFig.5showsatyPeofladderlmoldedPartswiththeladderholefeatUre,whieheanbeobtainedPerformingBooleansubFetionbetweentwoladdershaftfeatires,andafilledeornerastheassistantfeatire.ThereasoningsearehingProeessis,zstlyPerfongtherstjudglnent,iffeature115theladdershaft,thentheease1withladdershaft15obtained:thenPerinnlngtheseeondjudgrnent,ifthefeatUre2withladershaftPerforinsBooleansubtraetionwitheaseZ,andthePositionofitseenterline15atthesarneloeonwiththefonnerones,thentheladdershafteasewithladderholefeatire15obtained.Applyingthisreasongmethodeanaeauirehi1efieienev.DecausesearcningalongoneembranementneednottraverseallofeaseinthelibraJ/.lfthereasoningrulesandeasesaremature,thesimilareasewiththeshaPefeatureofthemoldedParteanbequleklyobtainedbyaPplyinghierarehiealsearehingmethodbasedontheProduetrule.Furthennore,duringtheProcessofeaseBasedPartingsuriteeautomatiedesign,theeasemodelneednotresembletheParntmodelentirely.ItonlyneedsthattheirmainshaPefeaturesaresimilarandthepartingsurfaeePosirionsrelativetooneofnshaPefeaturesofthetwomodelsaresIar.The517ltysolutionmodelofPartingsurfaeeautomatiegenerating15ProPosedhere.4SIMILARITYSOLUTIONMODELOFPARTINGSURFACEThePartingsurfaeeofaeasehasthesPeeialPositionrelationrelativetoaeertainfeane,AndtheserelationsrePresentalwaysonacerntainsurfaee,aeertainseetionorPassingthroughaeertainPointofthefeateete.TesePositionrelationseanbeexPressedasaZDvectof,thatisP=,n)whereL-BSPeeialPositionrelationBBBVeetorsymbolC-BPoiniofthefealewhjehthePrtingstlrfaeePassesiNormalveetorofPartingsuri(eeInotherwords,theParntingsuri(eeisrresentedbyanornalveetorandasPeeialPointPofthefeature.ForeachPartings+eofaeasemodel,veetorX15unebangeable.AsfarasaboardclikePartcasemodel15eoneemed,ifthePartingsurfaee15deternnedonthetoPfaee,thenPointP15theeenterPointofthisFce,and15thenonnalveetor.Firstly,findthePositionrelationPofthePartingsurfaeerelativetotheeasemodel,andthensubstitutethePositionrelationintothePartmodeltodetenmneitsPartlngsurfacePosition.ThesolvingproeesscanbeexPressedbythefollowingexPressionI(O(p)c=+(O(p)1)CPmwheresPiPartingsurfaeeC7BBBCasemodelOiPartmodel1,FiFeatUrelocatedinCmandPmresPectivelyThelenitemrePresentsthePartieularPositionofPartingsurfaeerelativetoacertainfeatureintheeasemodel,whiletherightonerePresentsthisrelationinthePartmodel.Usingtheir517larityweeandetenInethePaIntingsurfaceofPartmodel.AcordingtothecotrLPletyofthesFmeofthemoldedPts,thepartingsMfedesignlryincfudeseveralasPeCtsasfoUows.(l)TbePartingrface15Plane.ForthosePartswhosemainfeaeface15Plane,oneeanfinddbeetlypointPandveetor.SubstitutingthemintotheveetorexPressionX(P,n),thePartingsur+cePositionofthePartmoldeanbedetemuned.ForthePartshopinFig.6,thetoPface15seleetedastheParntingsu(eeaceordingtotheeasesinthe11bra.ThusPointP15thetoPfaeeeenterPointofthecubiealPartmodel,whilen15thenormalveetorbf5suree,andthese0valuescanbefoundmomthefeaturee,50thepositionofPartlngsurfaeecanbe50xvedeonvenientiy.(2)LadderPartingsu)ee.TheladderPartingfaeefeature15comlnonlyusedinsomemoldedPartsForthePartshownjnFig.7,.612.dTongmin,etal:Desi,teChnolofotinjectionmoldpartingsurfacobasedoncasesandowledgeoneneedfiqtlyfindthevectorsxl,l,ofeaehladderfaee,thentheladderPartingsurfaCeeanbeObtainedbyPerforungBooleanunionoPerationwiththePartingsurfaeeseontrolledtheseveetogsurfaee.IntheProcessofthePartingsurfaeeselecting,theknowledgelibrary,theeaseslibrB/,andthefoardBreasoningmethodareused.ThesimilaritysolutionmethodandBooleantooperatlonam9thechildpartingslRfacesareusedtogeneratethefinalPaZnt1ngsue.Fig.1015theZDilfustrationofPartingsur+1eofthisPartmodel.q3)ThePartingsurfaeoiscMrvedface.The501vlngProeess15eomPlieatedwhenthePartingfaee15eurvedfaee.TakethePartshowninFig.8asanexamPle,thekeyfealeParameters,suchastheradiusarc,thePOsitionofeenterlineanditsdistribution,sho-uldbefoundaeeordingtotheshaPeoftheeurvedface.Thenthesolvingmodeleanbeeonstlpcted.AftersubstitUtingtheseParab-etersintothesolvingmodel,thePartingIineeanbegenerated,andthenthePartingsurfaeealsoebeObtained.IfthePartingsurface15freeBwsurface,thesolvingmedelwilbemorecomPlieated(4)MoldedPartswitheomPlieatedsNe.ThemoreeomPlieatedPartmodel15theonethatitssimilareasedoesnotexistintheeaselibraryofPartingsurfaeePosition,andtheeasemodelresultedfromthereasoningProeedurealsohasasignifieantdiferenee.Thiskindofeasemodelshouldbesolveddireetlyatthistime,forthegeneralstrUetLireofthePart15eo+licated,butthePositionofPartinglinemaybesi7lar.+kethePartshowninFig.9asanexamPle,whichconsistsofseveralsFeefaors,there15notthe517lareaseintheeaselibrary.ButitsPartingfaeealsoeanbedesignedusingthecaseshowninFig.6,andthetoPfaceofthePartwouldbechosenasthePartingsurfacebeeausetheirPositionsofPngsurfaeeare517lar.IfthesimilaritysolutionProeeduredoesnotwork,thehumanBeomPuterinteraetionwillbechosentodetnnlnetheb0ip,surfaeeandtheresultwillbestored,thusthecaselibrBofth-PartlngsurtacePosltlonw1JIOeenncneQc6CONCLUSIONSTheknowleeandeaseBb=edreasoningdesignofmoldPrtlngsurface15toreasonthe517larcasewiththeoriginalPdexedbyshaPefeatureoftheparntandtodeterninethePositionofthemoldParntingseeemPIOyingthe517laritysolutionmethod.ThisProachhashighdesignefieieneyandreliabilityonthebasisoftheabundantknowleelibrandcaselibrary.ltalsohasseveralcharacteristicsasfollows:(I)ThehieraehiealeaselibrBofPartingsurfacePosition15eonsetedbedonthenumberofthefeaesintheease,whlchmxesthesearebF9rnoreeonvenient.(2)IndexedtheshaPefeane,ahierarchiealsearchingmethod15emPloyedbasedontheProduetrule.IthashighefieieneyandeanquieklyreasonouttheeasemodelwhiehmainshaPefeale15slinilartothemoldedPart.(3)Usingthe517larityoftheshaPefeaturesbetweentheeasemodelandthePartmodel,thesilaritysolutionmodelofPartingsud)ce15founded,whicheandetenInethePositionofthemeldPngsuri(cequieklyandexactiy.Asaresult,thequalityandtheefi