机器模具加工测量系统 英语翻译.docx

机器模具加工测量系统 英语翻译.docx

《机器模具加工测量系统 英语翻译.docx》由会员分享，可在线阅读，更多相关《机器模具加工测量系统 英语翻译.docx（20页珍藏版）》请在冰点文库上搜索。

机器模具加工测量系统英语翻译

AWeb-·basedon-·machinemouldmatchingandmeasurementsystembasedonCAD／CAM／CAIintegration

一个基于在·用机器制造配比的模子和根据CAD/CAM/CAI综合化的测量系统

Abstract：

ThepurposeofthisstudyistodevelopaWeb—basedon—machinemouldidentificationandmeasurementsystem．TheWeb—basedmouldidentificationsystemmatchesobtainedvisioninformationwithCADdatabase．DevelopedWeb．basedsystemistoexchangemessagesbetweenaserverandaclientbymakingofActiveXcontro1．andtheresultofmouldidentificationisshownonWeb—browseratremotesite．Foreffectivefeatureclassificationandextraction．thesignaturemethodisusedtomakemeaningfulinformationfromobtainedimagedata．Foron—machinemeasurementofthematchedmould．inspectiondatabaseisconstructedfromCADdatabaseusingdevelopedinspectionplanningmethods．Theresultsaresimulatedandanalyzedusingdevelopedsystemtoverifytheeffectivenessoftheproposedmethods．

Keywords：

E—manufacturing，CAD／CAM／CAIintegration，On—machinemeasurement，Mouldidentification，Visionsystem

doi：

10．1631／jzus．2006．A1474Documentcode：

ACLCnumber：

TP202

INTRODUCTION

Today，themanufacturingindustrymakesconstantefortinordertoconstructanE—manufacturingsystemusinginformationtechnologyforeffectiveproductioncontrol，andthen，itofferstheintegratedproductioncontroltotheworkers，theproductionmanagers，andtheconsumers，etc．（Lee，2003；Choi，2004）．Inthemouldmanufacturingindustry,theprocessmanagersoftenrequestpaperworktoknowtheinformationoftheprocessesonCNCmachinetoolsortheyvisittheshopfloor（Choetal．，2003）．Thoseineffectiveactivitiescausetheworker’scomplain，thedelayoftheprocesses，andthequalityproblems．Eventhepointofproductioncontrolusingthebarcodetechniquealsoburdenstheworkerontheshopfloor．Muto（2003）proposedtheproductioncontrolsystembyinstallingcamerasontheshopfloor,but，itonlyshowsthemachiningstatusofmachinetool．Thedetailedinformationcannotbeacquiredbyhismethod．Thoughthereverseengineeringsystemusing3Dvisioninformationcanbeusedforthepurpose，theamountofitsdatageneratedistoomuchtobedealtwithinarealtimeWeb-basedsystem（Choetal．，2000；Choi，2003）

Inthedevelopedsystemofthisresearch，aWebcameramonitorstheprocessingandtakesanimageofamachinedmould，featuresarerecognizedfromtheimage，andtheyarematchedwiththefeatureswhichareacquiredinCADdatabase，then，amouldcanbeidentified．Usingthesystem,themouldanditsprocessinginformationcanbesharedwithworkers。

Themoreappropriatedecisionmakingbythecontrolmanagerispossiblefortheeffectiveproduction，andtheremotecustomermonitorstheorderedmouldpartsvisuallythroughtheWeb．

Aftermouldidentificationprocessisperformed．On-machinemeasurementforthemouldcanbeperformedtochecktheaccuracyofthemachinedpartsbyrequestfromtheremotesite．Forthispurpose，inspectionplanningdatabaseisconstructedforeachmouldfromCADdatabase．

Eachfeatureisdecomposedintoitsconstituentgeometricelementssuchasplane，circle，etc．Then，aseriesoftasksareprocessedsuchasf11suitablenumberofmeasuringpoints，

（2）theirlocations，and（3）optimumprobingpath，tominimizemeasuringerrorsandtimes．Finally，theproposedmethodsaresimulatedusingdevelopedsystemtoverifytheireffectivenessinrealprocess．Fig．1illustratestheproposedsysteminthisresearch．

FEATURECLASSIFICAIT0NANDDATABASEC0NSTRUCTION

Databaseformouldmatching

Thedevelopedsysteminthisresearchconstructsadatabaseforthe3DCADmodelandmatchesthefeatureinformationfromvisualimagewiththedatabase．Thevisualimagerepresentsthetopviewfromtheorthogonaldirectiononthepartitionedsurfaceofthemouldcavity．ThedatabaseiscomprisedofthetopviewsoftheCADdatafilesofthemouldcavities．ThegeometricalobjectinformationisextractedusingUGAPI．anditisstoredinadatabaseaccordingtothepredefinedfeaturestandards．ThestandardfeatureshavetobesimplydefinedgeometricallyinvisualimagesandacquiredinaCADdatafileconveniently．ThestructureisrepresentedinFig．2．Theinformationofitsarea，perimeterandthecoordinatesofthecenterareusedinordertodifferentiateeachgeometricobjectinfeature．

ThestructureofthedatabaseisconstructedasanMDB（ManufacturingDatabase）inFig．3．ItconsistsofamainDBanditssub—DBscorespondingtomouldsrespectively．ThetypesofthefeatureandtheirnumbersarestoredinthemainDB．Ifthecombinationofthefeaturesofamouldisverydifierentfromthatofthetargetmould．itwouldbeignored．Thesub—DBsconsistoftheinformationofthearea．perimeterandthecoordinatesofthecenter，andareusedtomatchthecandidatemouldsselectedusingthemainDBwiththetargetmould．ThematchingprocesswiththefeatureinformationfromthevisualimageisperformedusingSQL．

DatabaseforOMM

ForthemeasuringprocessusingOMM，atouch—typeprobeisgenerallyused，whichperformspoint·-to·-pointmotionstogetdimensionaldataofthetargetsurface，usuallyonepointatatime（ChoandSeo，2002）．Althoughthetargetworkpieceisverycomplicatedandhasmanymachiningfeatures．itiscomposedofmanygeometricprimitivessuchasplanes，cylinders，andsculpturedsurfaces，etc．AsshowninFig．4（ChoandKim，1995；Choeta1．，1995；Mantylaeta1．，1994；Rogers，1994）．Also，whenusingatouchtypeprobe，theproberepeatsidenticalpoint-to-pointmotionstogetpointdataoftheprimitives。

Thegeometrictoleranceevaluation，suchasparallelism,squareness，etc．，canbedeterminedbymanipulatingtheobtainedpointdataset．Thus，itisnecessarytodecomposethefeaturesintoeasy-to-handleprimitivesforeffectiveinspectionplanning（ChangandWysk，1985）

AsshowninFig．5．thefeaturedecompositionprocedureiscomposedoftwosteps：

decompositioninto

（1）inspectionfeaturesand

（2）primitivesforlocalinspectionplanning．Inthefirststepofthedecompositionintoinspectionfeatures．theobjectisdecom．posedintothefeaturesforinspectionsuchaspockets．planesandislands，etc．（Fig．5a）．Inthisstep，thedecomposedfeaturesareverysimilartomachiningfeatures．AsshowninFig．5b．thefeaturesobtainedformthefirststepcanbedecomposedintothegeometricprimitivessuchasplanes，cylinders，etc．Sincetheinspectionprocessusing0MMwithatouch-type

probeperformspoint·-to·-pointmotionstogetthedimensionaldataofthetargetsurface．suchdecompositionprocessmakesiteasiertobuildaninspectionplanforcomplicatedobjects.

M0ULDIDENTIFICAT10NSYSTEM

Web-basedmatchingsystem

AWebserverandclientsystemisconstructedforthisresearch．andthesocketcommunicationisimplementedfortheeffectivemessageexchangebetweentheserverandtheclient（Jungeta1．2001）．AnActiveXcontrolismadeusingVC++forthesmoothandpromptconnectionwiththeserverandthemutualcommunication；itisdownloadedautomaticallytothefirstconnectedclient．

Thesocketcommunicationcarriesoutthefo1.1owingthreeeventsforthemouldidentificationre．questasshowninFig．6．First．theclientrequeststheservertosendtheimagebytheWebcamerainstalledonthemachine．Atiertheserversendstherequestedimagestoredinit，theclientdisplaystheimageontheclientprogrammadeofActiveX．Second．thecoordinatesofthecroppedportionbytheclientaresenttotheserver．TheserverperformstheimageprocessingofthevisionsysteminthecroppedimagebytheWebcamerausingthereceivedcoordinates，andthen，featureinformationiscreated．ThefeatureinformationismatchedwiththedatabaseconstructedusingUG．API．andthemostsimilarmouldisselected．Third．theserversendsouttheinformationoftheselectedmould，andtheclientdisplaystheinformation．Then，thesocketcommunicationends．

Visionsystemandimageprocessing

ThevisualimageforthemouldmatchingisshowninFig．7．Theedgedetectionalgorithmandtheedgeslimmingoperationareappliedtothesmoothedandenhancedimagetoacquirethefeatureinformation．Ifthelengthoftheperimeterofafeatureexceedsapredeterminedvalue．thefeatureisexcluded．Thesignaturemethod（AhnandCho，1996）whichcalculatesthecentroidofafeatureandthedistancetotheedgefromthecentroidisusedtocalculatethecoordinatesofthecenter，thelengthoftheperimeter，andthearea.

ThematchingprocessoftheimagedataacquiredbythevisionsystemandthedatabaseisshowninFig．8．Intheprocess，because3Dimagerequiresmuchmorecalculationthan2Dimage，the2Dimageisusedfortheimageacquisitionandprocessing．Preprocessingincludingthecleaningprocessfornoisereductionisperfclrmedonthe2DimagebytheWebcamera．andtheessentialinformationisacquiredf0rthepredeterminedfeatures．Thehighpassfilter．

1owpassfilter，blurringfilter，medianfilter，andthehistogramequalizerareusedforthepre-processing；thebluringfilter；thecombinationofbinaryprocessing；thecombinationofdilationanderosiontech．niquesareusedforthecleaningprocess．

TheextractedfeatureinformationisusedforthetwostagematchingprocesseswiththeCADdatabase．Thecorespondingmouldisselectedandtheextractedfeaturesarecomparedindetailtoconfirmasthetargetmould．Becausethisprocessreliesonmatching．notthedimensionalinformation，theresultcanbeobtainedrapidlyandeasilyusingthefeatureinformationoflowprecisionleve1.

Thesignaturemethodcanrepresenttheinfomationofthe2DboundarypointsaslDfunctionsasshowninFig．9．Itplotsthedistancetotheboundaryfromthecentroid．thenitcandescribethecharacteristicsofageometricalfigure．Inthisstudy．the2DboundaryinformationisrepresentedasthelDinformationforeachpredefinedfeature．Th