tionandself-healingwillbepresentedwithgreatfeaturesaswellaschallengesrelatedtozeroofintelligentbenefitsTheMachiningprocessmonitoringandcontrolisacoreconceptonwhichtobuildupthenewgenerationofflexibleself-opti-misingintelligentNCmachines.In-processmeasurementandprocessingoftheinformationprovidedbydedicatedsensorsinstalledinthemachine,enablesautonomousdecisionmakingbasedontheon-linediagnosisofthecorrectmachine,work-piece,toolandmachiningprocesscondition,leadingtoanincreasedmachinereliabilitytowardszerodefects,togetherwithhigherproductivityandefficiency.Indeed,themainsensingandprocessingtechniquesintheliterature35focuson0094-114X/$-seefrontmatterC2112008ElsevierLtd.Allrightsreserved.*Correspondingauthor.Tel.:+441612003804.E-mailaddress:s.mekidmanchester.ac.uk(S.Mekid).MechanismandMachineTheory44(2009)466476ContentslistsavailableatScienceDirectMechanismandMachineTheorydoi:10.1016/j.mechmachtheory.2008.03.006underverytightconditions1,2.Themachine-toolindustryisrespondingtoanumberofrequirements,e.g.e_commerce,just-in-time-productionandmostimportantlyzerodefectcomponent.Thisisfacilitatedbyintegratingnewmaterials,designconcepts,andcontrolmech-anismswhichenablemachinetoolsoperatingathigh-speedwithaccuraciesbelowthan5lm.Howevertheintegrationofhumanexperienceinmanufacturingtowardsflexibleandself-optimisingmachinesiswidelymissing.Thiscanbeachievedbyenhancingexistingcomputingtechnologiesandintegratingthemwithhumanknowledgeofdesign,automation,machin-ingandservicingintoe-manufacturing.Thenextgenerationwillbedescribedasnewintelligentreconfigurablemanufacturingsystemswhichrealisesadynamicfusionofhumanandmachineintelligence,manufacturingknowledgeandstate-of-the-artdesigntechniques.Thismayleadtolow-costself-optimisingintegratedmachines.Itwillencompassfault-tolerantadvancedpredictivemaintenancefacilitiesforproducinghigh-qualityerror-freeworkpiecesusingconventionalandadvancedmanufacturingprocesses.1.IntroductionComplexcomponentmachinedwithlengerequiredforthenewgenerationuctsandprocessesofferssubstantialhigherqualityandbetterreliability.variousaspectsofthenextgenerationofintelligentmachinetoolcentres.C2112008ElsevierLtd.Allrightsreserved.defectsisatopperformanceinmassproductionanditbecomesanewchal-machine-tools.Increasingtheprecisionandaccuracyofmachines,prod-toawiderangeofapplicationsfromultra-precisiontomassproductionwithrecentdevelopmentofultraprecisionmachinesisreachingnanometreprecisionBeyondintelligentmanufacturing:AnewgenerationofflexibleintelligentNCmachinesS.Mekida,*,P.Pruschekb,J.HernandezcaTheUniversityofManchester,SchoolofMechanical,AerospaceandCivilEngineering,ManchesterM601QD,UKbInstituteforControlEngineeringofMachineToolsandManufacturingUnits,UniversityofStuttgart,GermanycIDEKOTechnologicalCentre,ArriagaKalea,220870ElgoibarGipuzkoa,SpainarticleinfoArticlehistory:Received30November2006Receivedinrevisedform3March2008Accepted4March2008Availableonline29April2008abstractNewchallengesforintelligentreconfigurablemanufacturingsystemsareontheagendaforthenextgenerationofmachinetoolcentres.Zerodefectworkpiecesandjust-in-timepro-ductionaresomeoftheobjectivestobereachedforbetterqualityandhighperformanceproduction.Sustainabilityrequiresaholisticapproachtocovernotonlyflexibleintelligentmanufacturebutalsoproductandservicesactivities.Newroutesphilosophyofpossiblemachinearchitecturewithcharacteristicssuchashybridprocesseswithin-processinspec-journalhomepage:/locate/mechmtOntheotherhand,specialattentionhastobepaidtothelatterprocesscontrolstrategies(ACO).CharacteristicexamplesS.Mekidetal./MechanismandMachineTheory44(2009)466476467canbefoundat1519.Themainfunctionalityprovidedbysuchcontrolsystemsisthepost-processself-optimisationofprocessparameterset-up(i.e.feeds,depthsofcut,etc.),withtheobjectiveofset-uptimeminimisation,processknowledgemanagementandprocessoptimisation,towardsflexiblejust-in-timeproduction.Withthein-processmonitoringofprocessperformanceandthepost-processmeasurementoftheresultingpartquality,aknowledgebasedprocessmodelisusedtodeterminethenewoptimisedsetofcuttingparameters,enablingautonomousself-optimisation.Inthesameway,asapre-vioussteptooptimisation,ACOsystemsarealsoappliedtoselectthefirstprocessset-upfornewpartqualityandprocessrequirements.Therefore,ifaflexibleintelligentNCmachinetoolistobedeveloped,processknowledgebasedmodelsareacomponentofprimaryimportancetobeintegratedunderthemachinetoolcontrolarchitecture.Inadditiontotheadaptationofcontrolparametersaccordingtoprocessconditions,controlparametershavealsotobeoptimalduringhandling(includingchangingoperationsofworkpiecesandtools)andpositioningoperationsastheseoper-ationsaccounttypicallyformorethan50%oftheoveralloperatingtime.Earliermethodsforparameteroptimisationcon-centratedonthereductionofpositioningandsettlingtimesofthefeedaxisbytuningonlyafewbasiccontrolparameters(e.g.gainofthepositioncontrolloopandgainandresettimeofthevelocitycontrolloop).Withincreasedcom-putationalpower,optimisationmethodsasdescribedin20cannowbereinvestigatedfortheusewithawiderparametersetincludingtheparametersforaccelerationandjerklimitswhicharedirectlyinfluencingthevibrationsofanaxis.Ifthecharacteristicsofacontrolledaxisareknownbymeansofthevibrationbehaviour,anadequategenerationoftheprogrammedtrajectoriescanyieldafurtheroptimisation.Methodsforinputshaping49canbeusedtodesigntrajectoriesthatdonotexciteresonantfrequenciesofagivensystem.Hence,settlingtimesandthuspositioningtimescanbefurtherreduced.Concerningparameteroptimisationthroughself-learningparticularly,theinterestoftheso-calledmachinelearningap-proaches21willbeintroducedasthemainresearchtrendinprocessmonitoringandcontrolstrategiestowardstheintel-ligentmanufacturingsystem.2.ExpectedcharacteristicsofthenextgenerationTheexpectedcharacteristicsofthenextgenerationofmachinecentresaredescribedasfollows:(a)Integration:developmentofanintegratedmachinetoolbeingcapableofperformingbothconventionalandnon-con-ventionalprocessesinoneplatform.(b)Bi-directionaldataflow:definitionofabi-directionalprocesschainforunifieddatacommunicationexchangebetweenCAD,CAM,CNCandDrivesystems.(c)Processcontrolloop:developmentandCNCinte-grationofrobustandreliablereal-timestrategiesforthein-processtool,part,andprocessconditionmonitoringandcontrol.(d)Predictivemaintenance:specificationofaload-andsituation-dependentconditionmonitoringformachinecomponentsasabasisforself-reliantmachineoperation.Thiswillbefollowedbytheformulationofaself-organisingpredictivemain-tenanceschedulethatisbasedonself-andremotediagnosticsandcoversbothshortandlongtermaspects.(e)Autonomousoptimisation:developmentofaself-configuringself-optimisingcontrolsystemforautonomousmanufacturing,basedonthein-processmonitoring,characterisationandmanagementofprocessknowledge.Tofacilitatesuchcharacteristics,thefollow-ingtopicswillbenecessarytobeimplemented:(a)Todevelopanintegratedintelligentmachinecentrededicatedtoe-manufacturing.(b)Toinvestigateanddevelopfast,stableandstiffreconfigurablemachineswithhybridmachiningprocessestoprepareanewplatformforfuturemachine-tools.(c)Toinvestigateimplementationoftotalerrorcompensationandinsituinspectionfacility.monitoringstrategiesforpartconditionmonitoring(surfaceroughness,surfaceintegrityanddimensionalaccuracy),toolconditionmonitoring(theso-calledTCMforwearandbreakagedetection),processconditionmonitoring(chatteronsetandcollisiondetection)andmachinecomponentconditionmonitoringforpredictivemaintenancepurposes(rotarycompo-nentsandpartssubjecttofrictionsuchasguideways).Sincedirectandin-processmeasurementisnotgenerallypossibleduetotheaggressiveenvironmentinthecuttingzonesurroundings,themainresearcheffortoverthelastdecadesforpartandtoolmonitoringhasbeenfocusedonindirectmeasurementtechniques(processcondition-based),inwhichcuttingprocesscharacteristics(i.e.cuttingforcesandpower,vibrations,cuttingtemperature,acousticemission,etc.)aremeasuredinordertoindirectlyinferthepartandtoolcondition6,7.SensitivityofferedbyCNCinternalservosignalsfromopenarchitecturecontrollersisunderstudyaswell8,9,sincetheyenablethedevelopmentofmonitoringandcontrolstrategieswithouttheneedofinstallingadditionalsensorsinthemachine.Inthesameway,basedonthedataprovidedbyin-processmonitoring,autonomousself-optimisationcanbeperformedwiththeintegrationofprocesscontrolstrategiesintothemachinetoolcontrolarchitecture.Machiningprocesscontrolstrat-egiesareclassifiedintotwomaingroups5,namelyadaptivecontrolconstraint(ACC)andadaptivecontroloptimisation(ACO).IntheformerACCcontrolstrategies,aprocessvariable(i.e.cuttingforce)iskeptconstantandundercontrolthroughthereal-timein-processregulationofacuttingprocessparameter(i.e.cuttingfeed),withtheaimofincreasingprocessproduc-tivityandrepeatability.MainresearcheffortsonACCstrategiesfocusoncuttingforcecontrol1012andchattervibrationsuppression13,14.drawbacktodealwith.468S.Mekidetal./MechanismandMachineTheory44(2009)466476Indeed,flexiblemonitoringsystemsarerequiredundertheactualmarketrequirementsandthus,reliableprocessdiag-nosisisnecessaryunderdifferentcuttingconditions.Nowadays,acommonproblematicsharedbyconventionalprocessmonitoringapproachesforpartandtoolconditionmonitoringisthelackofreliabilityunderchangingcuttingconditionshencelimitingtheflexibilityofsuchautomationsystems3.Asacharacteristicexampleofthisproblematicforprocesscon-ditionbasedtoolconditionmonitoring(TCM),theprocessconditionisnotonlyinfluencedbychangesintoolcondition,butitisalsodirectlyaffectedbycuttingconditions.Furthermore,underdifferentcuttingconditions,differentwearmechanismscanbeactivatedonthetool,eachonehavingitsparticularimpactonprocessandpartcondition.Therefore,whensetting-upprocessmonitoringsystemsfornewcuttingconditions,previoustrialsforprocesssignaldatabaseretrievalarerequired4.Thesearecombinedtogetherwithskilledoperatorswiththenecessaryprocessknowledgeinordertointerpretchangesinprocessbehaviour(i.e.forces,vibrations,etc.)andset-upsuiteddetectionlimits.Additionally,flexibleprocessmonitoringequipmentsoftenrequiresadditionalsensorsthatcanfailandresultinunforeseendowntime.Asaresult,whenhighflex-ibilityisrequired,monitoringsystemsareusuallyswitched-offinindustry,anddirectpost-processmeasurementisper-formed,withthecorrespondingreliabilitylackinthemachinedpartquality.Dealingwithsuchaproblematic,model-basedprocessmonitoringandsensorfusionapproachesarepointedoutasthealternativeinordertogetreliableprocessconditiondiagnosis,withaclearresearcheffortoverlastyearsformachiningpro-cessessuchasturning2224,grinding4,25,26andmilling27.Ontheotherhand,theintegrationofhumanexperienceinmanufacturingiswidelymissingconcerningmachiningpro-cessoptimisation.Set-uptimereductioniscriticalwhenflexiblejust-in-timeproductionisrequired.Nowadays,set-up-timemainlydependsonprocessknowledgeconcentratedinskilledoperators,andthereisalackofsystematicmanagement,re-trieval,sharingandoptimisationofthatkeyknowledge.Furthermore,characterisationofprocessknowledgeanddevelop-mentofmodelsforautonomousprocessoptimisationarerequiredifset-uptimesaretobedrasticallyreduced.(d)Todevelopandproducenewmethodologiesandconceptsofautonomousmanufacturing,self-supervisionandself-diagnostic/tuning/healing.(e)Todevelopandintegratereal-timeprocesscontrollersintoopenCNCanddrivesystemarchitecture,takingthemachinefromanaxis-controlledsystemtoamachiningprocess-controlledself-reliantsystem,basedontheon-lineinformationprovidedbyrobustandreliablesensingtechniquesfortool,part,andmachiningprocessconditionmonitoring.(f)TodevelopandincorporateanextendibleandknowledgebasedCAMsystemcapableofrecognisingcomplexfeatures,performingself-learningbasedonin-processmonitoreddataprovidedbymachinecontrolloops,andautonomouslydeterminingtheoptimumtools/setsforgivenrequirementsofpartquality,machineproductivityandprocesseffi-ciency.Followingthee-manufacturingapproach,inasecondstep,CAMsystemscapableofsharingself-optimisedpro-cessknowledgebetweennetworkedmachinesaretobedeveloped.Aninterdisciplinaryapproachofmachine-toolbuildersinordertoachievetheseobjectivesbecomesnecessaryandin-cludescontrolmanufacturers,researchinstitutionsandpotentialend-users.Suchadevelopmentwillrealiseanumberofbreakthroughsinthefuture,e.g.(a)Delay-freecumzero-downtimeproduction:theproposede-manufacturingapproachwillseetheuseofelectronicservicesbasedonavailabledatafrommachinedprocesses,sensorsignals,andhumanexperiencethatisintegratedinazerodelay-timesystemtoenablemachineswithnearzero-downtimeandproductionthatmeetsuserrequirementswithzerodelaytime.(b)Self-reliantproduction:machineswillbeenabledtooperatewidelyautonomously.(c)Optimalproduction:self-configurationandself-optimisationwilleliminateproductionerrorsdowntothelimitationsofthein-processmeasurementdevices.3.ConceptsofintelligentandflexiblemachinesInFig.2,theauthorsproposeanewintegratedconceptforthenextgenerationofmachinetoolcentres.Basedontheknowledgeacquiredandthefeaturesextracted,theperformanceofcontrolsystemswillbeextendedtowardsself-controlledmanufacturingwiththeobjectivesofcost-effective,highquality,fault-tolerantandmoreflexiblesystemswithbetterpro-cesscapability.NewintelligentcontrolsystemshavetobedevelopedandintegratedwithopenarchitecturecontrollerssuchasOpenCNCC210orOSACA-basedCNCs.Inordertoallowanautomatederror-freeproductionwithnearzerodowntime,openinterfaces,learningcapabilities,self-tuningandself-adjustingmechanismsaswellassophisticatedmodel-basedpredictioninstrumentshavetobeimplementedattheselayers.Qualityinspectioncouldoperateinsituwithenvironmentalconditionstakenintoaccount.Forthefirsttime,theconceptofself-healingwithe-maintenancecouldbe