簡介：EFFICIENTPLANNINGOFSUBSTATIONAUTOMATIONSYSTEMCABLESTHANIKESAVANSIVANTHIANDJANPOLANDABBSWITZERLANDLTD,CORPORATERESEARCH,SEGELHOFSTRASSE1K,5405,BADEND¨ATTWIL,AARGAU,SWITZERLANDABSTRACTTHEMANUALSELECTIONANDASSIGNMENTOFAPPROPRIATECABLESTOTHEINTERCONNECTIONSBETWEENTHEDEVICESOFASUBSTATIONAUTOMATIONSYSTEMISAMAJORCOSTFACTORINSUBSTATIONAUTOMATIONSYSTEMDESIGNTHISPAPERDISCUSSESABOUTTHEMODELINGOFTHESUBSTATIONAUTOMATIONSYSTEMCABLEPLANNINGASANINTEGERLINEAROPTIMIZATIONPROBLEMTOGENERATEANEFFICIENTCABLEPLANFORSUBSTATIONAUTOMATIONSYSTEMS1INTRODUCTIONCABLINGBETWEENDIFFERENTDEVICESOFASUBSTATIONAUTOMATIONSYSTEMSASISAMAJORCOSTFACTORINTHESASDESIGNPROCESSUSUALLYCOMPUTERAIDEDDESIGNSOFTWAREISUSEDTOCREATETHEDESIGNTEMPLATESOFSASDEVICESANDTHEIRINTERCONNECTIONSTHEDESIGNTEMPLATESARETHENINSTANTIATEDINASASPROJECTANDTHECABLESAREMANUALLYASSIGNEDTOTHECONNECTIONSTHESELECTIONANDASSIGNMENTOFCABLESTOCONNECTIONSMUSTFOLLOWCERTAINENGINEERINGRULESTHISENGINEERINGPROCESSISUSUALLYTIMECONSUMINGANDCANCAUSEENGINEERINGERRORS,THEREBYINCREASINGTHEENGINEERINGCOSTAPPARENTLY,THESASCABLEPLANNINGISRELATEDTOTHEWELLKNOWNBINPACKINGPROBLEMTHESASCABLEPLANNINGCANBEFORMULATEDASANINTEGERLINEAROPTIMIZATIONPROBLEMWITHTHECABLEENGINEERINGRULESEXPRESSEDASASETOFLINEARCONSTRAINTSANDACOSTOBJECTIVEFORMINIMIZINGTHETOTALCABLECOSTTHISPAPERDESCRIBESTHEFORMULATIONOFSASCABLEPLANNINGPROBLEMASANINTEGERLINEAROPTIMIZATIONPROBLEMANDPRESENTSTHERESULTSFORSOMEREPRESENTATIVETESTCASESTOTHEBESTOFTHEAUTHORS’KNOWLEDGETHEWORKISTHEFIRSTOFTHEKINDTOSTUDYSASCABLEPLANNINGTHEPAPERISORGANIZEDASFOLLOWSSECTION2PRESENTSANOVERVIEWOFTHESASCABLEPLANNINGPROCESSSECTION3EXPRESSESTHESASCABLEPLANNINGPROBLEMASANINTEGERLINEAROPTIMIZATIONPROBLEMTHERESULTSOBTAINEDBYSOLVINGTHEOPTIMIZATIONPROBLEMUSINGSOMESOLVERSISPRESENTEDINSECTION4SECTION5DRAWSSOMECONCLUSIONSOFTHISWORK2SASCABLEPLANNINGTHESASCABLEPLANNINGBEGINSAFTERTHESYSTEMDESIGNPHASEOFASASPROJECTTHESASCABLEPLANNINGISATPRESENTDONEMANUALLYBYCOMPUTERAIDEDDESIGNTACHTERBERGANDJCBECKEDSCPAIOR2011,LNCS6697,PP210–214,2011CSPRINGERVERLAGBERLINHEIDELBERG2011EFFICIENTPLANNINGOFSUBSTATIONAUTOMATIONSYSTEMCABLES211TOTALNUMBEROFCONNECTIONS,ANDK{1,2,3,,M}212TSIVANTHIANDJPOLANDREPRESENTTHESETOFALLCABLETYPES,WHEREMISTHETOTALNUMBEROFCABLETYPESINASUBPROBLEMINACABLEINSTANCE,THERECANBEONEORMORECONNECTIONSANDWEREFERTOTHECONNECTIONWITHLOWESTINDEXAMONGALLCONNECTIONSINTHECABLEINSTANCEASTHELEADERANDTHEOTHERCONNECTIONSASTHEFOLLOWERSTHISIMPLIESTHATALLCONNECTIONSEXCEPTTHEFIRSTCONNECTIONINCCANEITHERBEALEADERORFOLLOWERMOREOVER,BASEDONTHESIGNALRULESASETOFCONNECTIONPAIRSXCANBEDERIVEDWHEREEACHI,?I∈XREPRESENTSTHECONNECTIONSIAND?ITHATMUSTNOTBEASSIGNEDTOTHESAMECABLELETˉCBETHESETOFCONNECTIONPAIRSI,?IWHEREI,?I∈C,I?I,I,?I/∈XWEINTRODUCETHEFOLLOWINGBINARYVARIABLEXI,?I,WHEREI,?I∈ˉC,WHICHWHENTRUEIMPLIESTHATCONNECTIONIISAFOLLOWEROFALEADER?I1CIIWHEREORXII??,,10,SIMILARLY,BASEDONTHECABLERULESASETOFCONNECTIONCABLEPAIRSYCANBEDERIVEDWHEREEACHI,J∈YIMPLIESTHATCABLETYPEJISNOTALLOWEDFORCONNECTIONILETˉKBETHESETOFCONNECTIONCABLEPAIRSI,J,WHEREI∈C,J∈K,I,J/∈YWEINTRODUCETHEFOLLOWINGBINARYVARIABLEYI,J,WHEREI,J∈ˉK,WHICHWHENTRUEIMPLIESTHATTHELEADERIISASSIGNEDTOANINSTANCEOFCABLETYPEJ2_,,,10KJIWHEREORXJI??TABLE1ILLUSTRATESALLBINARYVARIABLESCORRESPONDINGTOTHEEXAMPLESHOWNINFIGURE1FORTHECASEWITHTWOCABLETYPESK1ANDK2ITISASSUMEDTHATCONNECTIONSC1ANDC3CANNOTBEASSIGNEDTOTHESAMECABLEANDK1ISNOTANALLOWEDCABLETYPEFORCONNECTIONC3ASMENTIONEDBEFOREALLCONNECTIONSEXCEPTTHEFIRSTCONNECTION,WHICHMUSTBEALEADER,CANEITHERBEALEADERORFOLLOWERTHISISENSUREDBYTHEFOLLOWINGCONSTRAINT（3）CIYXKJIKJJICIIII?????????,1,,,,,ACONNECTIONWHICHISALEADERINACABLECANNOTBEAFOLLOWEROFALEADERINANOTHERCABLETHISISEXPRESSEDBYTHEFOLLOWINGCONSTRAINT（4）_,,,,,1_CIIXXCIIIIII??????ANIMPLICITCONSTRAINTOFTHECABLEPLANNINGPROBLEMISTHECAPACITYCONSTRAINTWHICHIMPLIESTHATTHENUMBEROFCONNECTIONSASSIGNEDTOACABLEMUSTBELESSTABLE1BINARYVARIABLESCORRESPONDINGTOFIGURE1EXAMPLE

簡介：附錄附錄B英文原文及翻譯英文原文及翻譯

簡介：英文翻譯系別自動化系專業(yè)自動化班級191003學(xué)生姓名周兵學(xué)號103658指導(dǎo)教師聶聰1引言萬向者往往是在當(dāng)代運用戰(zhàn)術(shù)導(dǎo)彈。他們應(yīng)該提供快速，準確的由目標檢測器產(chǎn)生的視軸誤差信號的跟蹤設(shè)在內(nèi)部萬向支架，在導(dǎo)引頭控制的要求更為嚴重的結(jié)局部分參與。性能結(jié)果丟失的距離不夠大因此降低了一個成功的截取的概率。一個兩自由度的（2DOF）的速率陀螺儀通常安裝在內(nèi)部萬向支架，并直接饋送慣性角速率，以扭矩裝置提供瞄準誤差跟蹤和穩(wěn)定反對基地運動1，2。后者是導(dǎo)彈的角和直線運動的參與過程的結(jié)果并通過機械裝置傳送到平衡環(huán)。準確的尋求穩(wěn)定的成像是至關(guān)重要的減少圖像涂抹,足夠的目標獲取,進而影響分割和跟蹤。此外,小質(zhì)量的平衡增加干擾的平衡環(huán)的導(dǎo)彈加速度。在戰(zhàn)術(shù)導(dǎo)彈子系統(tǒng)的包裝嚴重受容積和空氣動力學(xué)限制，最終規(guī)定的可操作性。萬向求職者通常定位在導(dǎo)彈的前末端。不是很少的大小導(dǎo)引頭及其配套制度決定的形狀導(dǎo)彈的前部尖端。在這種情況下，形狀笨重，更激烈成為產(chǎn)生沖擊波的降低導(dǎo)彈的性能。導(dǎo)引頭可以通過減少從萬向部件卸下速率陀螺和使用一個捷聯(lián)式結(jié)構(gòu)。然而，這種方法要求內(nèi)部萬向支架角速率相對于該的估計彈體。的相對角速度分化萬向節(jié)，并進一步匹配濾波，以減少噪音一直在一個線性化方法中使用3，以穩(wěn)定的單一軸萬向成像導(dǎo)引頭的運動不安。不正確由圖像分割算法的輸出被忽視?；？刂埔咽褂?下的假設(shè)的非耦合相同俯仰和偏航通道再次單軸萬向?qū)б^和評價對代表命令信號。需要提出的控制律的第一和第二時間導(dǎo)數(shù)的計算命令信號，以及萬向架的完美測量角位移和相對彈體率。本文提出延長上述配方，以應(yīng)付具有偏航和俯仰控制成像的動態(tài)導(dǎo)引頭。該方法是基于模型的非線性與在其隨時間變化的慣性導(dǎo)引頭使用動態(tài)擴展卡爾曼濾波（EKF），針對的估計相對角位移的平衡環(huán)。此外，圖像序列分析，假設(shè)目標分割已經(jīng)解決，并在其質(zhì)心位置的有噪聲估計圖像平面是可供在光學(xué)方面解決流21估計的視覺反饋的扭矩裝置。該方法是通過評估與脫靶的統(tǒng)計評估通過蒙特卡羅模擬閉環(huán)包括導(dǎo)引頭的控制和戰(zhàn)術(shù)十字形的動態(tài)模型導(dǎo)彈。該導(dǎo)彈是由純

簡介：PSEUDOPOLARBASEDESTIMATIONOFLARGETRANSLATIONSROTATIONSANDSCALINGSINIMAGESYOSIKELLERAMIRAVERBUCHMOSHEISRAELIDEPARTMENTOFMATHEMATICSDEPARTMENTOFCOMPUTERSCIENCEDEPARTMENTOFCOMPUTERSCIENCEYALEUNIVRSITYTELAVIVUNIVERSITYTECHNIONINSTITUTEOFTECHNOLOGYNEWHAVEN,CT,USATELAVIV,ISRAELHAIFA,ISRAELYOSIKELLERYALEEDUABSTRACTONEOFTHEMAJORCHALLENGESRELATEDTOIMAGEREGISTRATIONISTHEESTIMATIONOFLARGEMOTIONSWITHOUTPRIORKNOWLEDGETHISPAPERPRESENTSAFOURIERBASEDAPPROACHTHATESTIMATESLARGETRANSLATION,SCALEANDROTATIONMOTIONSTHEALGORITHMUSESTHEPSEUDOPOLARTRANSFORMTOACHIEVESUBSTANTIALIMPROVEDAPPROXIMATIONSOFTHEPOLARANDLOGPOLARFOURIERTRANSFORMSOFANIMAGETHUS,ROTATIONANDSCALECHANGESAREREDUCEDTOTRANSLATIONSWHICHAREESTIMATEDUSINGPHASECORRELATIONBYUTILIZINGTHEPSEUDOPOLARGRIDWEINCREASETHEPERFORMANCEACCURACY,SPEED,ROBUSTNESSOFTHEREGISTRATIONALGORITHMSSCALESUPTO4ANDARBITRARYROTATIONANGLESCANBEROBUSTLYRECOVERED,COMPAREDTOAMAXIMUMSCALINGOF2RECOVEREDBYTHECURRENTSTATEOFTHEARTALGORITHMSTHEALGORITHMUTILIZESONLY1DFFTCALCULATIONSWHOSEOVERALLCOMPLEXITYISSIGNIFICANTLYLOWERTHANPRIORWORKSEXPERIMENTALRESULTSDEMONSTRATETHEAPPLICABILITYOFTHESEALGORITHMS1INTRODUCTIONIMAGEREGISTRATIONPLAYSAVITALROLEINMANYIMAGEPROCESSINGAPPLICATIONSSUCHASVIDEOCOMPRESSION1,VIDEOENHANCEMENT2ANDSCENEREPRESENTATION3TONAMEAFEWTHISPROBLEMWASANALYZEDUSINGVARIOUSCOMPUTATIONALTECHNIQUES,SUCHASPIXELDOMAINGRADIENTMETHODS2,CORRELATIONTECHNIQUES15ANDDISCRETEFOURIERDFTDOMAINALGORITHMS6,11GRADIENTMETHODSBASEDIMAGEREGISTRATIONALGORITHMSARECONSIDEREDTOBETHESTATEOFTHEARTTHEYMAYFAILUNLESSTHETWOIMAGESAREMISALIGNEDBYONLYAMODERATEMOTIONFOURIERBASEDSCHEMES,WHICHAREABLETOESTIMATERELATIVELYLARGEROTATION,SCALINGANDTRANSLATION,AREOFTENUSEDASBOOTSTRAPFORMOREACCURATEGRADIENTMETHODSTHEBASICNOTIONRELATEDTOFOURIERBASEDSCHEMESISTHESHIFTPROPERTY18OFTHEFOURIERTRANSFORMWHICHALLOWSROBUSTESTIMATIONOFTRANSLATIONSUSINGTHENORMALIZEDPHASECORRELATIONALGORITHM6,9,10HENCE,INORDERTOACCOUNTFORROTATIONSANDSCALING,THEIMAGEISTRANSFORMEDINTOAPOLARORLOGPOLARFOURIERGRIDREFERREDTOASTHEFOURIERMELLINTRANSFORMROTATIONSANDSCALINGAREREDUCEDTOTRANSLATIONSINTHESEREPRESENTATIONSANDCANBEESTIMATEDUSINGPHASECORRELATIONINTHISPAPERWEPROPOSETOITERATIVELYESTIMATETHEPOLARANDLOGPOLARDFTUSINGTHEPSEUDOPOLARFFTPPFFT19THERESULTINGALGORITHMISABLETOROBUSTLYREGISTERIMAGESROTATEDBYARBITRARYANGLESANDSCALEDUPTOAFACTOROF4ITSHOULDBENOTEDTHATTHEMAXIMUMSCALEFACTORRECOVEREDIN11AND16WAS20AND18,RESPECTIVELYINPARTICULAR,THEPROPOSEDALGORITHMDOESNOTRESULTTOINTERPOLATIONINEITHERSPATIALORFOURIERDOMAINONLY1DFFTOPERATIONSAREUSED,MAKINGITMUCHFASTERANDESPECIALLYSUITEDFORREALTIMEAPPLICATIONSTHERESTOFPAPERISORGANIZEDASFOLLOWSPRIORRESULTSRELATEDTOFFTBASEDIMAGEREGISTRATIONAREGIVENINSECTION2,WHILETHEPROPOSEDALGORITHM,ISPRESENTEDINSECTION3EXPERIMENTALRESULTSAREDISCUSSEDINSECTION4ANDFINALCONCLUSIONSAREGIVENINSECTION52PREVIOUSRELATEDWORK21TRANSLATIONESTIMATIONTHEBASISOFTHEFOURIERBASEDMOTIONESTIMATIONISTHESHIFTPROPERTY18OFTHEFOURIERTRANSFORMDENOTEBYFFFX,YG,BFΩX,ΩY1THEFOURIERTRANSFORMOFFX,YTHEN,FFFX￠X,Y￠YGBFΩX,ΩYEJΩX￠XΩY￠Y2EQUATION2CANBEUSEDFORTHEESTIMATIONOFIMAGETRANSLATION6,10ASSUMETHEIMAGESI1X,YANDI2X,YHAVESOMEOVERLAPTHATI1X￠X,Y￠YI2X,Y3PROCEEDINGSOFTHEIEEEWORKSHOPONMOTIONANDVIDEOCOMPUTINGWACV/MOTION’050769522718/052000IEEEROTATIONANDTRANSLATIONESTIMATIONALGORITHMOPERATESASFOLLOWS1LETM1,L1ANDM2,L2BETHESIZESOFI1I,JANDI2I,J,RESPECTIVELYTHEN,ATITERATIONN0,I1I,JANDI2I,JAREZEROPADDEDSUCHTHATM1L1M2L22K,K2Z122THEMAGNITUDESMPP1?ΘI,RJ￠ANDMPP2?ΘI,RJ￠OFTHEPPFFTSOFIN1I,JANDI2I,JARECALCULATED,RESPECTIVELY3THEPOLARDFTS,MAGNITUDESCMPOLAR1?ΘI,RJ￠ANDCMPOLAR2?ΘI,RJ￠OFIN1I,JANDI2I,JARESUBSTITUTEDBYMPP1?ΘI,RJ￠ANDMPP2?ΘI,RJ￠RESPECTIVELY4THETRANSLATIONALONGTHE?ΘAXISOFMPP1?ΘI,RJ￠ANDMPP2?ΘI,RJ￠ISESTIMATEDUSINGPHASECORRELATIONTHERESULTISDENOTEDBY￠ΘN5LETΘNBETHEACCUMULATEDROTATIONANGLEESTIMATEDATITERATIONNΘN,NXI0￠ΘIΘN?1￠ΘNTHEN,THEINPUTIMAGEI1I,JISROTATEDBYΘNAROUNDTHECENTEROFTHEIMAGEUSINGTHEFFTBASEDIMAGEROTATIONALGORITHMDESCRIBEDIN4THISROTATIONSCHEMEISACCURATEANDFASTSINCEONLY1DFFTOPERATIONSAREUSEDIN11Θ,RI01ΘΘN,R,N1,THEROTATIONCANBECONDUCTEDAROUNDANYPIXELWERECOMMENDTOUSETHECENTRALPIXELOFI1I,JSUCHTHATTHEBOUNDINGRECTANGULAROFTHEROTATEDIMAGEWILLBEASSMALLASPOSSIBLE6STEPS25AREREITERATEDUNTILTHEANGULARREFINEMENTTERM￠ΘNISSMALLERTHANAPREDEFINEDTHRESHOLDΕΘ,IEJ￠ΘNJ0THEPOLARAXISISAPPROXIMATEDUSINGTHESAMEPROCEDUREASINSECTION31,WHILETHERADIALAXISISAPPROXIMATEDUSINGNEARESTNEIGHBORINTERPOLATION4THERELATIVETRANSLATIONBETWEENCMLOG?POLAR1I,JANDCMLOG?POLAR2I,JISRECOVEREDBYA2DPHASECORRELATIONONTHE?ΘAND?RAXES5LET￠ΘNAND￠RNBETHEROTATIONANGLEANDTHESCALINGVALUEESTIMATEDATITERATIONN,RESPECTIVELYTHEN,THEINPUTIMAGEI1X,YISROTATEDAROUNDTHECENTEROFTHEIMAGE4ANDTHENSCALEDUSINGDFTDOMAINZEROPADDINGIN11Θ,RI01ΘΘN,R￠RN16WHEREΘNNXI0￠ΘIΘN?1￠ΘNRNNYI0￠RIRN?1￠￠RN17PROCEEDINGSOFTHEIEEEWORKSHOPONMOTIONANDVIDEOCOMPUTINGWACV/MOTION’050769522718/052000IEEE

簡介：LEARNINGMULTIROBOTJOINTACTIONPLANSFROMSIMULTANEOUSTASKEXECUTIONDEMONSTRATIONSMURILOFERNANDESMARTINSDEPTOFELECANDELECTRONICENGINEERINGIMPERIALCOLLEGELONDONLONDON,UKMURILOIEEEORGYIANNISDEMIRISDEPTOFELECANDELECTRONICENGINEERINGIMPERIALCOLLEGELONDONLONDON,UKYDEMIRISIMPERIALACUKABSTRACTTHECENTRALPROBLEMOFDESIGNINGINTELLIGENTROBOTSYSTEMSWHICHLEARNBYDEMONSTRATIONSOFDESIREDBEHAVIOURHASBEENLARGELYSTUDIEDWITHINTHEFIELDOFROBOTICSNUMEROUSARCHITECTURESFORACTIONRECOGNITIONANDPREDICTIONOFINTENTOFASINGLETEACHERHAVEBEENPROPOSEDHOWEVER,LITTLEWORKHASBEENDONEADDRESSINGHOWAGROUPOFROBOTSCANLEARNBYSIMULTANEOUSDEMONSTRATIONSOFMULTIPLETEACHERSTHISPAPERCONTRIBUTESANOVELAPPROACHFORLEARNINGMULTIROBOTJOINTACTIONPLANSFROMUNLABELLEDDATATHEROBOTSFIRSTLYLEARNTHEDEMONSTRATEDSEQUENCEOFINDIVIDUALACTIONSUSINGTHEHAMMERARCHITECTURESUBSEQUENTLY,THEGROUPBEHAVIOURISSEGMENTEDOVERTIMEANDSPACEBYAPPLYINGASPATIOTEMPORALCLUSTERINGALGORITHMTHEEXPERIMENTALRESULTS,INWHICHHUMANSTELEOPERATEDREALROBOTSDURINGASEARCHANDRESCUETASKDEPLOYMENT,SUCCESSFULLYDEMONSTRATEDTHEEFFICACYOFCOMBININGACTIONRECOGNITIONATINDIVIDUALLEVELWITHGROUPBEHAVIOURSEGMENTATION,SPOTTINGTHEEXACTMOMENTWHENROBOTSMUSTFORMCOALITIONSTOACHIEVETHEGOAL,THUSYIELDINGREASONABLEGENERATIONOFMULTIROBOTJOINTACTIONPLANSCATEGORIESANDSUBJECTDESCRIPTORSI29ARTIFICIALINTELLIGENCEROBOTICSGENERALTERMSALGORITHMS,DESIGN,EXPERIMENTATIONKEYWORDSLEARNINGBYDEMONSTRATION,MULTIROBOTSYSTEMS,SPECTRALCLUSTERING1INTRODUCTIONASUBSTANTIALAMOUNTOFSTUDIESINMULTIROBOTSYSTEMSMRSADDRESSESTHEPOTENTIALAPPLICATIONSOFENGAGINGMULTIPLEROBOTSTOCOLLABORATIVELYDEPLOYCOMPLEXTASKSSUCHASSEARCHANDRESCUE,DISTRIBUTEDMAPPINGANDEXPLORATIONOFUNKNOWNENVIRONMENTS,ASWELLASHAZARDOUSTASKSANDFORAGING–FORANOVERVIEWOFTHEFIELD,SEE13DESIGNINGDISTRIBUTEDINTELLIGENTSYSTEMS,SUCHASMRS,ISAPROFITABLECITEASLEARNINGMULTIROBOTJOINTACTIONPLANSFROMSIMULTANEOUSTASKEXECUTIONDEMONSTRATIONS,MFMARTINS,YDEMIRIS,PROCOF9THINTCONFONAUTONOMOUSAGENTSANDMULTIAGENTSYSTEMSAAMAS2010,VANDERHOEK,KAMINKA,LESPéRANCE,LUCKANDSENEDS,MAY,10–14,2010,TORONTO,CANADA,PP?COPYRIGHTC?2010,INTERNATIONALFOUNDATIONFORAUTONOMOUSAGENTSANDMULTIAGENTSYSTEMSWWWIFAAMASORGALLRIGHTSRESERVEDFIGURE1THEP3ATMOBILEROBOTSUSEDINTHISPAPER,EQUIPPEDWITHONBOARDCOMPUTERS,CAMERAS,LASERANDSONARRANGESENSORSTECHNOLOGYWHICHBRINGSBENEFITSSUCHASFLEXIBILITY,REDUNDANCYANDROBUSTNESS,AMONGOTHERSSIMILARLY,ASUBSTANTIALAMOUNTOFSTUDIESHAVEPROPOSEDNUMEROUSAPPROACHESTOROBOTLEARNINGBYDEMONSTRATIONLBD–FORACOMPREHENSIVEREVIEW,SEE1EQUIPPINGROBOTSWITHTHEABILITYTOUNDERSTANDTHECONTEXTINWHICHTHEYINTERACTWITHOUTTHENEEDOFCONFIGURINGORPROGRAMMINGTHEROBOTSISANEXTREMELYDESIREDFEATUREREGARDINGLBD,THEMETHODSWHICHHAVEBEENPROPOSEDAREMOSTLYFOCUSSEDONASINGLETEACHER,SINGLEROBOTSCENARIOIN7,ASINGLEROBOTLEARNTASEQUENCEOFACTIONSDEMONSTRATEDBYASINGLETEACHERIN12,THEAUTHORSPRESENTEDANAPPROACHWHEREAHUMANACTEDBOTHASATEACHERANDCOLLABORATORTOAROBOTTHEROBOTWASABLETOMATCHTHEPREDICTEDRESULTANTSTATEOFTHEHUMAN’SMOVEMENTSTOTHEOBSERVEDSTATEOFTHEENVIRONMENTBASEDONITSUNDERLYINGCAPABILITIESASUPERVISEDLEARNINGMETHODWASPRESENTEDIN4USINGGAUSSIANMIXTUREMODELS,INWHICHAFOURLEGGEDROBOTWASTELEOPERATEDDURINGANAVIGATIONTASKFEWSTUDIESADDRESSEDTHEPREDICTIONOFINTENTINADVERSARIALMULTIAGENTSCENARIOS,SUCHASTHEWORKOF3,INWHICHGROUPMANOEUVRESCOULDBEPREDICTEDBASEDUPONEXISTINGMODELSOFGROUPFORMATIONINTHEWORKOF5,MULTIPLEHUMANOIDROBOTSREQUESTEDATEACHER’SDEMONSTRATIONWHENFACINGUNFAMILIARSTATESIN14,THEPROBLEMOFEXTRACTINGGROUPBEHAVIOURFROMOBSERVEDCOORDINATEDMANOEUVRESOFMULTIPLEAGENTSALONGTIMEWASADDRESSEDBYUSINGACLUSTERINGALGORITHMTHEMETHODPRESENTEDIN9ALLOWEDASINGLEROBOTTOPREDICTTHEINTENTIONSOF2HUMANSBASEDONSPATIOTEMPORALRELATIONSHIPSHOWEVER,THECHALLENGEOFDESIGNINGANMRSSYSTEMINWHICHMULTIPLEROBOTSLEARNGROUPBEHAVIOURBYOBSERVATION931931938I1I2INF1F2FNSTATESATTM1M2MNPREDICTIONVERIFICATIONATT1PREDICTIONVERIFICATIONATT1PREDICTIONVERIFICATIONATT1P1P2PNFIGURE3DIAGRAMATICSTATEMENTOFTHEHAMMERARCHITECTUREBASEDONSTATEST,MULTIPLEINVERSEMODELSI1TOINCOMPUTEMOTORCOMMANDSM1TOMN,WITHWHICHTHECORRESPONDINGFORWARDMODELSF1TOFNFORMPREDICTIONSREGARDINGTHENEXTSTATEST1P1TOPNWHICHAREVERIFIEDATST1MAYPERFORMCERTAINACTIONSSEQUENTIALLYORSIMULTANEOUSLYRESULTINGINACOMBINATIONOFACTIONS,WHILETHEROBOTHASACCESSTOTHEJOYSTICKCOMMANDSONLYINORDERTORECOGNISEACTIONSFROMOBSERVEDDATAANDMANOEUVRECOMMANDS,THISPAPERMAKESUSEOFTHEHIERARCHICALATTENTIVEMULTIPLEMODELSFOREXECUTIONANDRECOGNITIONHAMMERARCHITECTURE7,WHICHHASBEENPROVENTOWORKVERYWELLWHENAPPLIEDTODISTINCTROBOTSCENARIOSHAMMERISBASEDUPONTHECONCEPTSOFMULTIPLEHIERARCHICALLYCONNECTEDINVERSEFORWARDMODELSINTHISARCHITECTURE,ANINVERSEMODELHASASINPUTSTHEOBSERVEDSTATEOFTHEENVIRONMENTANDTHETARGETGOALS,ANDITSOUTPUTSARETHEMOTORCOMMANDSREQUIREDTOACHIEVEORMAINTAINTHETARGETGOALSONTHEOTHERHAND,FORWARDMODELSHAVEASINPUTSTHEOBSERVEDSTATEANDMOTORCOMMANDS,ANDTHEOUTPUTISAPREDICTIONOFTHENEXTSTATEOFTHEENVIRONMENTASILLUSTRATEDINFIG3,EACHINVERSEFORWARDPAIRRESULTSINAHYPOTHESISBYSIMULATINGTHEEXECUTIONOFAPRIMITIVEBEHAVIOUR,ANDTHENTHEPREDICTEDSTATEISCOMPAREDTOTHEOBSERVEDSTATETOCOMPUTEACONFIDENCEVALUETHISVALUEREPRESENTSHOWCORRECTTHATHYPOTHESISIS,THUSDETERMININGWHICHROBOTPRIMITIVEBEHAVIOURWOULDRESULTINTHEMOSTSIMILAROUTCOMETOTHEOBSERVEDACTION3SYSTEMIMPLEMENTATIONTHEMRLBDAPPROACHPROPOSEDINTHISPAPERISDEMONSTRATEDUSINGTHEAFOREMENTIONEDPLATFORMFORROBOTTELEOPERATION,WHICHCONSISTSINACLIENT/SERVERSOFTWAREWRITTENINCTOCONTROLTHEP3ATROBOTSFIG1UTILISEDINTHEEXPERIMENTS,ASWELLASANIMPLEMENTATIONOFTHEHAMMERARCHITECTUREFORACTIONRECOGNITIONANDAMATLABIMPLEMENTATIONOFTHESCALGORITHMSIMILARTOTHEONEPRESENTEDIN14ANOVERVIEWOFTHETELEOPERATIONPLATFORMCANBESEENINFIG4THESERVERSOFTWARECOMPRISESTHEROBOTCOGNITIVECAPABILITIESANDRESIDESONTHEROBOT’SONBOARDCOMPUTERTHESERVERISRESPONSIBLEFORACQUIRINGTHESENSORDATAANDSENDINGMOTORCOMMANDSTOTHEROBOT,WHEREASTHECLIENTSOFTWARERUNSONAREMOTECOMPUTERANDSERVESASTHEINTERFACEBETWEENTHEHUMANOPERATORANDTHEROBOT31THEROBOTCOGNITIVECAPABILITIESWITHINTHEROBOTCOGNITIVECAPABILITIESBLOCK,THESERVERCOMMUNICATESWITHTHEROBOTHARDWAREBYUSINGTHEWELLKNOWNROBOTCONTROLINTERFACEPLAYER6,WHICHISANETWORKSERVERTHATWORKSASAHARDWAREABSTRACTIONLAYERTOINTERFACEHUMANROBOTINTERFACEROBOTCOGNITIVECAPABILITIESWIFINETWORKJOYSTICKVISUALISATIONROBOTCONTROLENVIRONMENTPERCEPTIONPLAYERSERVERLOGGINGROBOTHARDWAREPLANEXTRACTIONACTIONRECOGNITIONHAMMERGROUPBEHAVIOURSEGMENTATIONMULTIROBOTPLANFIGURE4OVERVIEWOFTHETELEOPERATIONPLATFORMDEVELOPEDINTHISPAPERWITHAVARIETYOFROBOTICHARDWAREINITIALLY,THEINTERNALODOMETRYSENSORSAREREADTHISDATAPROVIDESTHECURRENTROBOT’SPOSE,WHICHISUPDATEDASTHEROBOTMOVESAROUNDANDUSEDASTHEGROUNDTRUTHPOSEFORCALCULATINGOBJECTS’POSEANDBUILDINGTHE2DMAPOFTHEENVIRONMENTODOMETRYSENSORSAREKNOWNFORINHERENTLYADDINGINCREMENTALERRORSANDHENCELEADTOINACCURATEPOSEESTIMATIONSBUTNEVERTHELESS,ITISSHOWNLATERONINSECTION5THATTHISINACCURACYWASIMMATERIALTOTHERESULTSTHEIMAGECAPTURED320X240PIXELS,COLOUREDFROMTHEROBOT’SCAMERAAT30FRAMESPERSECONDISCOMPRESSEDUSINGTHEJPEGALGORITHMANDSENTTOTHECLIENTSOFTWAREOVERATCP/IPCONNECTIONUSINGTHEWIFINETWORKADDITIONALLY,THEIMAGEISALSOUSEDTORECOGNISEOBJECTSBASEDUPONAKNOWNOBJECTSDATABASE,USINGTHEAPPROACHPRESENTEDIN15THISALGORITHMCONSISTSINDETECTINGTHEPOSECARTESIANCOORDINATESINTHE3DSPACE,PLUSROTATIONONTHERESPECTIVEAXESOFUNIQUEMARKERSTHEKNOWNOBJECTSDATABASECOMPRISESASETOFUNIQUEMARKERSANDTHEOBJECTTHATEACHMARKERISATTACHEDTO,ANDALSOOFFSETVALUESTOCOMPUTETHEPOSEOFTHEOBJECTBASEDUPONTHEDETECTEDMARKER’SPOSEASHORTMEMORYALGORITHM,BASEDUPONCONFIDENCELEVELS,WASALSOIMPLEMENTEDTOENHANCETHEOBJECTRECOGNITIONTHEOBJECT’SPOSEISTRACKEDFORAPPROXIMATELY3SECONDSAFTERITHASLASTBEENSEENTHISAPPROACHWASFOUNDEXTREMELYUSEFULDURINGTHEEXPERIMENTS,ASTHECOMPUTERVISIONALGORITHMCANNOTDETECTMARKERSFROMDISTANCESGREATERTHAN2METRESANDOCCLUSIONISLIKELYTOHAPPENINREALAPPLICATIONSTHESICKLMS200LASERRANGESCANNERPROVIDESMILLIMETREACCURACYDISTANCEMEASUREMENTSFROMUPTO80METRES,RANGINGFROM0DEGREESRIGHTHANDSIDEOFTHEROBOTTO180DEGREESLEFTHANDSIDEINADDITION,16SONARRANGESENSORS,PLACEDINARINGCONFIGURATIONONTHEROBOT,RETRIEVEMODERATELYACCURATEDISTANCEMEASUREMENTSFROM01TO5METRESANDA30DEGREEFIELDOFVIEWEACHDESPITETHELACKOFPRECISION,THESONARSENSORSPLAYAFUNDAMENTALROLEINTHEOVERALLOUTCOMEOFTHETELEOPERATIONPLATFORMASTHEHUMANOPERATORHASLIMITEDPERCEPTIONOFTHEENVIRONMENT,PARTICULARMANOEUVRESMAINLYWHENREVERSINGTHEROBOTMAYBEPOTENTIALLYDANGEROUSANDRESULTINACOLLISIONTHUS,OBSTACLEAVOIDANCEISACHIEVEDBYUSINGANIMPLEMENTATIONBASEDUPONTHEWELLKNOWNALGORITHMVFHVECTORFIELDHISTOGRAM2HOWEVER,THEHUMANOPERATORISABLETOINHIBITTHESONARREADINGSASDESIRED,FEATUREWHICHISUSEFULWHENPUSHINGOBJECTS,PASSINGTHROUGHNARROWGAPSAND933

簡介：556IEEETRANSACTIONSONCONTROLSYSTEMSTECHNOLOGY,VOL10,NO4,JULY2002LINEOFSIGHTRATEESTIMATIONANDLINEARIZINGCONTROLOFANIMAGINGSEEKERINATACTICALMISSILEGUIDEDBYPROPORTIONALNAVIGATIONJACQUESWALDMANN,MEMBER,IEEEABSTRACTACCELERATIONCOMMANDSINMISSILESGUIDEDBYPROPORTIONALNAVIGATIONREQUIRETHEMEASUREMENTOFLINEOFSIGHTLOSRATEITISOFTENOBTAINEDBYFILTERINGTHEOUTPUTOFATWODEGREEOFFREEDOM2DOFRATEGYROMOUNTEDONTHEINNERGIMBALOFTHESEEKERTHISPAPERDESCRIBESTHEMODELINGOFANIMAGINGSEEKERANDTHEFORMULATIONOFANEXTENDEDKALMANFILTEREKFFORTHEESTIMATIONOFLOSRATEFROMMEASUREMENTSOFRELATIVEANGULARDISPLACEMENTBETWEENSEEKERGIMBALSANDALOWCOSTSTRAPDOWNINERTIALUNITTHEAPPROACHAIMSATCIRCUMVENTINGTHENEEDFORTHERATEGYROONTHESEEKERALINEARIZINGFEEDBACKCONTROLLAWFORDECOUPLINGMISSILEMOTIONFROMTHATOFTHESEEKERISPROPOSEDBASEDONTHEFILTERMODELANDITSESTIMATESADDITIONALLY,THECONTROLLAWUSESVISUALINFORMATIONFROMTHEIMAGESEQUENCEFORTARGETTRACKINGSEEKERDYNAMICSANDCONTROLARETHENINTEGRATEDINTOADYNAMICMODELOFACRUCIFORMMISSILEEQUIPPEDWITHCANARDSANDROLLERONSANDGUIDEDBYPROPORTIONALNAVIGATIONINTHREEDIMENSIONAL3DINTERCEPTIONTASKSMONTECARLOSIMULATIONISEMPLOYEDTOEVALUATETHEOVERALLSYSTEMACCURACYSUBJECTTODIFFERENTINITIALCONDITIONSLATERALANDHEADONENGAGEMENTSANDTHEIMPACTOFROLLINGMOTIONDURINGHIGHMANEUVERSONMISSDISTANCETHEVALIDATIONMODELINCLUDESNOISEINTHEVARIOUSSENSORS,COUPLEDINERTIAOFTHESEEKERGIMBALS,SIGNALSATURATIONATVARIOUSSUBSYSTEMS,OPTICALGEOMETRICDISTORTION,ANDTARGETSEGMENTATIONERRORSINTHEIMAGEPLANEINITIALENGAGEMENTGEOMETRYANDROLLRATEDAMPINGATHIGHINCIDENCEANGLESHAVEBEENOBSERVEDTOHAVEASIGNIFICANTIMPACTONMISSDISTANCEINDEXTERMSIMAGESEQUENCEANALYSIS,KALMANFILTERING,MACHINEVISION,MISSILEGUIDANCE,NONLINEARESTIMATIONANDCONTROL,OPTICALDISTORTION,POINTINGSYSTEMSIINTRODUCTIONGIMBALLEDSEEKERSAREOFTENUTILIZEDINCONTEMPORARYTACTICALMISSILESTHEYSHOULDPROVIDERAPIDANDACCURATETRACKINGOFBORESIGHTERRORSIGNALSGENERATEDBYTHETARGETDETECTORLOCATEDINTHEINNERGIMBALTHEDEMANDSONSEEKERCONTROLBECOMEMORESEVEREATTHEENDGAMEPORTIONOFTHEENGAGEMENTINADEQUATEPERFORMANCERESULTSINLARGEMISSDISTANCESANDTHUSREDUCESTHEPROBABILITYOFASUCCESSFULINTERCEPTIONATWODEGREEOFFREEDOM2DOFRATEGYROISUSUALLYMOUNTEDONTHEINNERGIMBALANDFEEDSINERTIALANGULARRATEDIRECTLYTOTHETORQUERSTOPROVIDEBORESIGHTERRORTRACKINGANDSTABILIZATIONAGAINSTBASEMOTION1,2THELATTERISACONSEQUENCEMANUSCRIPTRECEIVEDDECEMBER11,2000REVISEDNOVEMBER9,2001MANUSCRIPTRECEIVEDINFINALFORMFEBRUARY22,2002RECOMMENDEDBYASSOCIATEEDITORSBANDATHEAUTHORISWITHTHECENTROTéCNICOAEROESPACIAL,INSTITUTOTECNOLóGICODEAERONáUTICA,DEPARTMENTOFSYSTEMSANDCONTROL,12228900S?OJOSéDOSCAMPOSSP,BRAZILEMAILJACQUESELEITACTABRPUBLISHERITEMIDENTIFIERS1063653602053563OFTHEMISSILEANGULARANDLINEARMOTIONDURINGTHEENGAGEMENTANDISTRANSMITTEDTOTHEGIMBALSBYMECHANICALMEANSACCURATESTABILIZATIONOFIMAGINGSEEKERSISCRITICALTOREDUCEIMAGESMEARINGWHICHINTURNIMPACTSADEQUATETARGETACQUISITION,SEGMENTATION,ANDTRACKINGADDITIONALLY,MINORMASSUNBALANCESADDTOTHEDISTURBANCESACTINGUPONTHEGIMBALSASTHEMISSILESUFFERSACCELERATIONSTHEPACKAGINGOFSUBSYSTEMSINTACTICALMISSILESISSERIOUSLYAFFECTEDBYVOLUMEANDAERODYNAMICCONSTRAINTSTHATULTIMATELYDICTATEMANEUVERABILITYGIMBALLEDSEEKERSAREUSUALLYPOSITIONEDATTHEFRONTTIPOFTHEMISSILENOTRARELYTHESIZEOFTHESEEKERANDITSSUPPORTINGSYSTEMSDICTATESTHESHAPEOFTHEFRONTTIPOFTHEMISSILEINSUCHCASES,THEBULKIERTHESHAPE,THEMOREINTENSEBECOMETHEGENERATEDSHOCKWAVESWHICHDEGRADEMISSILEPERFORMANCESEEKERVOLUMECANBEREDUCEDBYREMOVINGTHERATEGYROFROMTHEGIMBALLEDASSEMBLYANDUSINGASTRAPDOWNCONFIGURATIONHOWEVER,THEAPPROACHCALLSFORTHEESTIMATIONOFTHEINNERGIMBALANGULARRATERELATIVETOTHEMISSILEBODYDIFFERENTIATIONOFTHERELATIVEANGULARRATEOFTHEGIMBALSANDFURTHERMATCHEDFILTERINGTOREDUCENOISEHASBEENUSEDINALINEARIZEDAPPROACH3TOSTABILIZEASINGLEAXISGIMBALLEDIMAGINGSEEKERDISTURBEDBYMISSILEMOTIONINCORRECTOUTPUTBYTHEIMAGESEGMENTATIONALGORITHMWASNEGLECTEDSLIDINGMODECONTROLHASBEENUSED4UNDERTHEASSUMPTIONOFUNCOUPLEDIDENTICALPITCHANDYAWCHANNELSAGAINWITHASINGLEAXISGIMBALLEDSEEKERANDEVALUATEDAGAINSTAREPRESENTATIVECOMMANDSIGNALTHEPROPOSEDCONTROLLAWREQUIREDTHECOMPUTATIONOFFIRSTANDSECONDTIMEDERIVATIVESOFTHECOMMANDSIGNALASWELLASPERFECTMEASUREMENTSOFGIMBALANGULARDISPLACEMENTANDRATERELATIVETOTHEMISSILEBODYTHISPAPERPROPOSESTOEXTENDTHEABOVEFORMULATIONTOCOPEWITHTHEDYNAMICSOFAYAWANDPITCHCONTROLLEDIMAGINGSEEKERTHEAPPROACHISBASEDONMODELINGTHENONLINEARSEEKERDYNAMICSWITHITSTIMEVARYINGINERTIAFORUSEINANEXTENDEDKALMANFILTEREKF,AIMINGATTHEESTIMATIONOFRELATIVEANGULARDISPLACEMENTOFTHEGIMBALSFURTHERMORE,IMAGESEQUENCEANALYSISASSUMINGTHATTARGETSEGMENTATIONHASBEENSOLVEDANDANOISYESTIMATEOFITSCENTROIDLOCATIONINTHEIMAGEPLANEISAVAILABLEISADDRESSEDINTERMSOFOPTICALFLOW21ESTIMATIONFORVISUALFEEDBACKTOTHETORQUERSTHEAPPROACHISEVALUATEDBYASSESSINGTHEMISSDISTANCESTATISTICSVIAMONTECARLOSIMULATIONOFTHECLOSEDLOOPCOMPRISINGSEEKERCONTROLANDTHEDYNAMICMODELOFATACTICALCRUCIFORMMISSILETHEMISSILEISGUIDEDBYPUREPROPORTIONALNAVIGATIONINTHREEDIMENSIONAL3DENGAGEMENTSAGAINSTONENONMANEUVERINGTARGET10636536/021700?2002IEEE558IEEETRANSACTIONSONCONTROLSYSTEMSTECHNOLOGY,VOL10,NO4,JULY2002DYNAMICSOFTHELASTTWOCOMPONENTSIN1BWHENEXCITEDBYTORQUEFROMTHEACTUATORSTHEFIRSTCOMPONENTREFERSTOTHEREACTIONTORQUEOFTHEMISSILEBODYACTINGUPONTHEOUTERGIMBALANDHENCEISNOTCONSIDEREDINTHEENSUINGMODELONONEHAND,TORQUECOMPONENTISAPPLIEDTOTHEOUTERGIMBALALONGTHEDIRECTIONANDAFFECTSTHEANGULARMOMENTUMCOMPONENTOFBOTHGIMBALSGIVENBYTORQUECOMPONENT,ONTHEOTHERHAND,ISAPPLIEDTOTHEINNERGIMBALALONGTHEDIRECTIONBYANACTUATORLOCATEDINSUCHWAYATTHEOUTERGIMBALTHATITSACTIONISPERPENDICULARTOTHEREFORE,ONLYAFFECTSTHEANGULARMOMENTUMCOMPONENTOFTHEINNERGIMBALALONGDIRECTION,GIVENBY4SUBSTITUTING2–4IN1BPRODUCESTHEFOLLOWING5A5BWHICHYIELDSAFTERSOMEALGEBRAICMANIPULATION6A6B6C6D6E6F6G6H6I6J6KTHEDYNAMICMODELOFSEEKERMOTIONRELATIVETOTHENBECOMES7AAND7BATTHEBOTTOMOFTHEPAGEANDTHEDYNAMICCOUPLINGARISINGFROMTHEINERTIAPRODUCTSBECOMESAPPARENTTORQUERDYNAMICSISMODELEDBY8WHERE,ARECURRENTSIGNALSAPPLIEDTOEACHTORQUERAND,ARECONSTANTGAINSTHECURRENTSIGNALSTOTHETORQUERSMUSTDRIVETHESEEKER,AIMINGATBASEMOTIONSTABILIZATIONANDTARGETTRACKINGINERTIAMOMENTSANDPRODUCTSOFBOTHGIMBALSANDELECTROOPTICALPAYLOADAREKNOWNALONGTORQUERAXESPRIORTOSEEKERASSEMBLYTHETORQUERAXESAREALIGNEDWITHTHECOORDINATEFRAMETHEREFORE,THEINERTIAPARAMETERSOFTHEINNERGIMBALASSEMBLYANDITSPAYLOADRELATIVETOTHEFRAMEVARYINTIMEDURINGSEEKEROPERATIONDUETOTHEOCCURRENCEOFRELATIVEMOTIONINELEVATION,GIVENBYANDTHEINERTIAMOMENTSANDPRODUCTSOFTHEINNERGIMBALANDELECTROOPTICALPAYLOADINTHEFRAME,ALONGWITHTHERESPECTIVETIMERATES,ARECOMPUTEDAS9EQUATIONS6–9COMPOSETHEDYNAMICMODELRELATINGTHEINPUTDRIVINGTHETORQUERSTOTHEOUTPUT,WHICHISTHESEEKERMOTIONRELATIVETOTHECOORDINATEFRAMEINORDERTOSTABILIZETHESEEKERININERTIALSPACEANDTRACKTHETARGETWHILEPERFORMINGPROPORTIONALNAVIGATION,THEINERTIALLOSRATEFROMSEEKERTOTARGETHASTOBEESTIMATEDFROMAVAILABLEMISSILEANGULARRATEANDRELATIVEGIMBALANGLEMEASUREMENTSINORDERTOCIRCUMVENTTHENEEDFORA2DOFRATEGYROMOUNTEDONTHEINNERGIMBALSECTIONIIIDESCRIBESTHEFORMULATIONOFANEKFFORTHISPURPOSEIIILOSRATEESTIMATIONVIAEKFTHEAVAILABLEMEASUREMENTSFORLOSRATEESTIMATIONARETHEOUTERGIMBALANGLERELATIVETOTHEMISSILEBODY,INNERGIMBAL7A7B

簡介：從同步任務(wù)執(zhí)行演示中學(xué)習(xí)多個機器人聯(lián)合行動計劃從同步任務(wù)執(zhí)行演示中學(xué)習(xí)多個機器人聯(lián)合行動計劃MURILOIEEEORGMURILOIEEEORGELECELEC與電子工程系，穆里羅費爾南德斯馬丁斯，英國倫敦帝國學(xué)倫敦，英與電子工程系，穆里羅費爾南德斯馬丁斯，英國倫敦帝國學(xué)倫敦，英國摘要摘要設(shè)計智能機器人通過行為示范已經(jīng)很大程度上影響了機器人系統(tǒng)的核心。許多架構(gòu)的認識和預(yù)測提出了一個教師的意圖。無論如何，很少的工作被完成訪問如何使一組機器人能夠在許多教師同時地示范中學(xué)習(xí)。本文有助于學(xué)習(xí)多個機器人聯(lián)合行動計劃不存在的數(shù)據(jù)。個人行為的機器人首先學(xué)習(xí)錘子架構(gòu)，隨后,運用時空聚類算法將行為分割在時間和空間上。根據(jù)實驗結(jié)果表明，人類遠程操作機器人在搜索和營救任務(wù)布置旗艦成功的示范了個人水平結(jié)合性行為識別和團體行為分割的功效，測定準確的時刻和讓機器人必須聯(lián)合實現(xiàn)預(yù)期的目標，因此生產(chǎn)一代合理的多功能機器人聯(lián)合行動計劃。分類和主題描述符號【人造的智力】機器人概述算法，設(shè)計，實驗關(guān)鍵詞關(guān)鍵詞從示范中學(xué)習(xí)，多功能機器人系統(tǒng)，光譜的采集11引言引言對多功能機器人系統(tǒng)研究的本質(zhì)發(fā)表演說，潛在的應(yīng)用程序保證大多數(shù)機器人能合作地部署復(fù)雜的任務(wù)，像搜索和營救。分配地圖和探索陌生的環(huán)境，有危險的任務(wù)和覓食一樣對于領(lǐng)域的一個綜述，看【13】，設(shè)計分散式的智能系統(tǒng)，比如，MPS，是賺錢的科技，他帶來的益處比如靈活性，裁員和穩(wěn)健性。本節(jié)還介紹錘架構(gòu)7和14提出的SC算法的實現(xiàn)是如何被利用來解決動作識別和群體行為的分割問題，其次，在第4部分描述了正是多功能機器人計劃的產(chǎn)生實驗性的測試已完成，第5部分分析得到的結(jié)果，最后，第6部分給出了結(jié)論和進一步的工作。2系統(tǒng)設(shè)計問題這篇文章的MOLBD體系結(jié)構(gòu)計劃是基于機器人遙控平臺，被【8】和【16】的工作所啟迪的設(shè)計，還有LBD體系結(jié)構(gòu)在【7】【9】所呈現(xiàn)的。一些MRS的設(shè)計包括了在這個研究區(qū)域的普遍問題，尤其是機器人遙控系統(tǒng)帶來了幾個核心的（重要的）設(shè)計問題，在接下來的章節(jié)會論述到。21人類VS機器人感知中心遙控平臺通常提供在機器人附著的遠程環(huán)境中受限制的直覺，然而，依賴于應(yīng)用和環(huán)境，人類可以以全面的。不受限制的觀察的這樣一種方式戰(zhàn)略性地被安置是可行的。第一被說明的設(shè)計問題是人類VS機器人幾種的知覺人被允許觀察自己的感覺世界或者他們應(yīng)該有自己的看法僅限于機器人介導(dǎo)的數(shù)據(jù)。。但先前的表現(xiàn)導(dǎo)致了簡化的系統(tǒng)，上述的MRS潛在運用難免會落后，執(zhí)行這項工作的遙控平臺因此基于對環(huán)境的受限制的知覺，為人類提空了機器人可以通過他的傳感器獲得的本地的相同的遙遠的知覺（人類被放置在機器人有知覺的鞋里）。22人類行為的觀察設(shè)計一個遙控操作平臺的另一個關(guān)鍵問題是如何定義相關(guān)的命令發(fā)送給機器人。人類的行為是不能被機器人直接觀察到的。盡管人類是“放置在機器人的感知中心一個機器人只能訪問它的遙控機器人的動作指令，而不是人類的動作一圖。如圖2所示。

簡介：三維空間攔截的前置追蹤變結(jié)構(gòu)制導(dǎo)律三維空間攔截的前置追蹤變結(jié)構(gòu)制導(dǎo)律葛連正,沈毅,高云峰,趙立軍哈爾濱工業(yè)大學(xué)控制科學(xué)與工程系，黑龍江哈爾濱150001摘要摘要為了解決導(dǎo)引頭探測由于高速運動所引起的干擾問題，提出了空間攔截的前置攔截方式。在建立了前置追蹤導(dǎo)引方式的三維制導(dǎo)模型的基礎(chǔ)上，對機動目標攔截基于李亞普諾夫穩(wěn)定性分析方法設(shè)計了一種前置追蹤非線性變結(jié)構(gòu)制導(dǎo)律。前置追蹤制導(dǎo)律將攔截器導(dǎo)引到目標軌道的前方進行攔截，要求攔截器的速度小于目標導(dǎo)彈的速度。攔截器和目標導(dǎo)彈彈道攔截的三維數(shù)字仿真驗證了制導(dǎo)模型和制導(dǎo)律的正確性。關(guān)鍵詞前置追蹤；三維制導(dǎo)模型；非線性變結(jié)構(gòu)；李亞普諾夫定理；制導(dǎo)律1引言引言在攔截戰(zhàn)術(shù)彈道導(dǎo)彈的攔截，多用來探測目標的紅外導(dǎo)引頭。然而，檢測精度往往是由于氣動加熱而退化1。為了解決氣動燒蝕問題，最近已開發(fā)的前置追蹤（HP）制導(dǎo)律攔截導(dǎo)彈，它的位置在對其飛行軌跡的目標摧毀目標2。利用該制導(dǎo)律，攔截器可以飛相同的方向與目標在一個較低的速度擊中目標。相比于正面接觸，低速度達到減少能源消耗。HP的指導(dǎo)方法是文獻中的進一步改進。相對運動模型可以被視為兩個垂直通道和制導(dǎo)問題每一個平面的問題。前置追蹤變結(jié)構(gòu)制導(dǎo)律進行了基于平面的模型。然而，由于實際導(dǎo)彈攔截發(fā)生在在三維空間中，一個三維的前置追蹤指導(dǎo)方法在實際中是比較有用的。各種經(jīng)典制導(dǎo)方法已檢查的三維制導(dǎo)攔截以來實施的三維純比例導(dǎo)引律由艾德勒提出的起源5。參考文獻611。已開發(fā)的三維制導(dǎo)模型，給出了基于李雅普諾夫穩(wěn)定性理論指導(dǎo)法。這些制導(dǎo)律只適宜迎面攔截，攔截方式和運動學(xué)模型不同于HP的指導(dǎo)方法。作為一個直觀的強大的控制技術(shù)，滑模變結(jié)構(gòu)控制1215一直在用各種指導(dǎo)應(yīng)用用來解決大的建模誤差和不確定性的非線性1??Ω?2COSCOS?COSCOS?COSCOSSINCOSTANCOSSIN?COSSIN??3COSSIN?COSSIN?COS?SINSINSIN?SIN?4SINTANCOSSIN?COSSIN?COSSIN?SIN?COSCOSSINCOSTAN?COSSIN?COSSIN??5COSSIN?COSSIN?COSSINSINSIN?SIN?SINTANCOSSIN?COSSIN?COS6SIN?SIN和VT分別是攔截器速度矢量和目標速度矢量。Ω1是LOS的視線角速度矢量。AYT和AZT分別是假定上的目標機動加速度和偏航機動加速度。AYM和AZM分別是俯仰機動加速度和攔截器的偏航機動加速度。前置追蹤制導(dǎo)律要求攔截器的速度低于目標，所以速度比定義為7N1為了達到目標，在攔截點R0不僅是必需的，但也需要目標在方向上攔截飛行器，因此,8LIM→00LIM→00,9LIM→00LIM→00指導(dǎo)法的目的是使前置追蹤的攔截器的達到這個點，這是限制的公式。（8）（9）。因此，攔截器導(dǎo)角ΘM和MΦ需要與目標的鉛角度相對瞄準線,1012

簡介：1一種尋的制導(dǎo)導(dǎo)彈模型參考變結(jié)構(gòu)自動駕駛儀的設(shè)計摘要設(shè)計某尋的導(dǎo)彈的自動駕駛儀回路,使導(dǎo)彈控制系統(tǒng)在正確響應(yīng)制導(dǎo)指令的同時,對彈體氣動力參數(shù)變化、量測噪聲等具有很好的抑制作用。將模型參考自適應(yīng)控制方法與變結(jié)構(gòu)控制方法相結(jié)合,為某型導(dǎo)彈設(shè)計了結(jié)構(gòu)簡單、實現(xiàn)方便的模型參考變結(jié)構(gòu)控制系統(tǒng)。仿真結(jié)構(gòu)表明,模型參考變結(jié)構(gòu)自動駕駛儀不僅能準確傳遞制導(dǎo)指令,而且具有很好的魯棒性,能有效地抑制氣動力浮動、量測噪聲等干擾因素。關(guān)鍵詞模型參考變結(jié)構(gòu),變結(jié)構(gòu)控制,自動駕駛儀導(dǎo)彈控制系統(tǒng)的任務(wù)如下第一個是穩(wěn)定彈體，使彈體具有適當(dāng)?shù)淖枘?，第二個是要正確引導(dǎo)轉(zhuǎn)移命令使舵偏轉(zhuǎn)和改變其性能最終目標是改變速度的大小和方向，并迫使導(dǎo)彈準確命中目標。然而，在導(dǎo)彈飛行期間，彈體的模型具有一定的不確定性，因為導(dǎo)彈的質(zhì)量，速度，和測量噪聲具有多樣性。此外，該導(dǎo)彈動力學(xué)在本質(zhì)上是高度非線性的。經(jīng)典控制理論和適應(yīng)性控制理論適合于線性發(fā)電站不能給出可靠的設(shè)計導(dǎo)彈自動駕駛儀?？勺兘Y(jié)構(gòu)控制系統(tǒng)理論在本質(zhì)上是一個控制理論的非線性系統(tǒng)，并且根據(jù)系統(tǒng)多樣性的現(xiàn)狀改變其結(jié)構(gòu)所以它能比常規(guī)的控制理論更有效的控制。此外可變結(jié)構(gòu)控制的滑動模式對于干擾更加穩(wěn)定，許多研究表明，可變結(jié)構(gòu)控制方法適用于導(dǎo)彈控制系統(tǒng)的設(shè)計。模型參考適應(yīng)性控制系統(tǒng)是一個很好的控制裝置，用于參數(shù)變化緩慢的線性發(fā)電站。自從控制發(fā)電廠和參考模型直接進行比較，適應(yīng)的速度高，控制器可以很容易地實現(xiàn)，但該模型參考適應(yīng)性控制只適合連續(xù)系統(tǒng)的模型是可以肯定的，并可能當(dāng)在有干擾，噪音和未建模動態(tài)的時候有不穩(wěn)定的現(xiàn)象。本文為了自導(dǎo)引導(dǎo)彈結(jié)合了可變結(jié)構(gòu)控制的模型參考適應(yīng)性控制和設(shè)計MRVS的自動駕駛儀。1模型參考變結(jié)構(gòu)系統(tǒng)設(shè)計由于ROLL穩(wěn)定的導(dǎo)彈系統(tǒng)的特性本文僅討論導(dǎo)彈的單輸入系統(tǒng)。302222212121???????SBUUBBXAAXAAPMMMM（8）將UP變?yōu)橄旅娴男问剑碝MPUKXKXKEKEKU?????24132211不等式為????????????????????????????00}{0}{00242222212121312121111122222121112111SUBKUBBSXBKXAAACASXBKXAAACASEBKEACASEBKEACAMMMMMMMMMMM（9）然后不等式8將成立，即UP會滿足變結(jié)構(gòu)控制的達成條件。如果B0，然后當(dāng)K1,K2,K3,K4,KM得到以下值，即????????????????????????????????????SGNMAXSGNMAXSGNMAXSGNMAXSGNMAX2222121242121111132212221111SBBBKSBAAAACKSBAAAACKSBACAKSBACAKMMMMMMMMMM（10）公式（9）將全部成立。為了抑制震動，一個飽和函數(shù)將取代開環(huán)函數(shù)SGNS。2一些尋的導(dǎo)彈自動駕駛儀設(shè)計一般來說，彈體是弱阻尼，所以螺距角速率的反饋通常被用來增加阻尼反饋循環(huán)也可以使從引導(dǎo)命令的傳輸系數(shù)的對過載的變化越小越好。本文中使用的內(nèi)部循環(huán)作為發(fā)電站設(shè)計MRVS的自動駕駛儀。彈體的傳遞函數(shù)通常被表示為其中N是過載，也就是彈體的輸出，U為引導(dǎo)命令的電廠輸入。為了使用MRVS，電廠的傳遞函數(shù)被首先改變?yōu)闋顟B(tài)空間模型。導(dǎo)彈本身的傳遞函數(shù)通常表示為

簡介：JOURNALOFCONTROLTHEORYANDAPPLICATIONS20075183–88DOI101007/S1176800552586NEUROFUZZYGENERALIZEDPREDICTIVECONTROLOFBOILERSTEAMTEMPERATUREXIANGJIELIU1,JIZHENLIU1,PINGGUAN21DEPARTMENTOFAUTOMATION,NORTHCHINAELECTRICPOWERUNIVERSITY,BEIJING102206,CHINA2DEPARTMENTOFAUTOMATION,BEIJINGINSTITUTEOFMACHINERY,BEIJING100085,CHINAABSTRACTPOWERPLANTSARENONLINEARANDUNCERTAINCOMPLEXSYSTEMSRELIABLECONTROLOFSUPERHEATEDSTEAMTEMPERATUREISNECESSARYTOENSUREHIGHEFFICIENCYANDHIGHLOADFOLLOWINGCAPABILITYINTHEOPERATIONOFMODERNPOWERPLANTANONLINEARGENERALIZEDPREDICTIVECONTROLLERBASEDONNEUROFUZZYNETWORKNFGPCISPROPOSEDINTHISPAPERTHEPROPOSEDNONLINEARCONTROLLERISAPPLIEDTOCONTROLTHESUPERHEATEDSTEAMTEMPERATUREOFA200MWPOWERPLANTFROMTHEEXPERIMENTSONTHEPLANTANDTHESIMULATIONOFTHEPLANT,MUCHBETTERPERFORMANCETHANTHETRADITIONALCONTROLLERISOBTAINEDKEYWORDSNEUROFUZZYNETWORKSGENERALIZEDPREDICTIVECONTROLSUPERHEATEDSTEAMTEMPERATURE1INTRODUCTIONCONTINUOUSPROCESSINPOWERPLANTANDPOWERSTATIONARECOMPLEXSYSTEMSCHARACTERIZEDBYNONLINEARITY,UNCERTAINTYANDLOADDISTURBANCE1,2THESUPERHEATERISANIMPORTANTPARTOFTHESTEAMGENERATIONPROCESSINTHEBOILERTURBINESYSTEM,WHERESTEAMISSUPERHEATEDBEFOREENTERINGTHETURBINETHATDRIVESTHEGENERATORCONTROLLINGSUPERHEATEDSTEAMTEMPERATUREISNOTONLYTECHNICALLYCHALLENGING,BUTALSOECONOMICALLYIMPORTANT3FROMFIG1,THESTEAMGENERATEDFROMTHEBOILERDRUMPASSESTHROUGHTHELOWTEMPERATURESUPERHEATERBEFOREITENTERSTHERADIANTTYPEPLATENSUPERHEATERWATERISSPRAYEDONTOTHESTEAMTOCONTROLTHESUPERHEATEDSTEAMTEMPERATUREINBOTHTHELOWANDHIGHTEMPERATURESUPERHEATERSPROPERCONTROLOFTHESUPERHEATEDSTEAMTEMPERATUREISEXTREMELYIMPORTANTTOENSURETHEOVERALLEFFICIENCYANDSAFETYOFTHEPOWERPLANTITISUNDESIRABLETHATTHESTEAMTEMPERATUREISTOOHIGH,ASITCANDAMAGETHESUPERHEATERANDTHEHIGHPRESSURETURBINE,ORTOOLOW,ASITWILLLOWERTHEEFFICIENCYOFTHEPOWERPLANTITISALSOIMPORTANTTOREDUCETHETEMPERATUREFLUCTUATIONSINSIDETHESUPERHEATER,ASITHELPSTOMINIMIZEMECHANICALSTRESSTHATCAUSESMICROCRACKSINTHEUNIT,INORDERTOPROLONGTHELIFEOFTHEUNITANDTOREDUCEMAINTENANCECOSTSASTHEGPCISDERIVEDBYMINIMIZINGTHESEFLUCTUATIONS,ITISAMONGSTTHECONTROLLERSTHATAREMOSTSUITABLEFORACHIEVINGTHISGOALTHEMULTIVARIABLEMULTISTEPADAPTIVEREGULATORHASBEENAPPLIEDTOCONTROLTHESUPERHEATEDSTEAMTEMPERATUREINA150T/HBOILER3,ANDGENERALIZEDPREDICTIVECONTROLWASPROPOSEDTOCONTROLTHESTEAMTEMPERATURE4ANONLINEARLONGRANGEPREDICTIVECONTROLLERBASEDONNEURALNETWORKSISDEVELOPEDIN5TOCONTROLTHEMAINSTEAMTEMPERATUREANDPRESSURE,ANDTHEREHEATEDSTEAMTEMPERATUREATSEVERALOPERATINGLEVELSTHECONTROLOFTHEMAINSTEAMPRESSUREANDTEMPERATUREBASEDONANONLINEARMODELTHATCONSISTSOFNONLINEARSTATICCONSTANTSANDLINEARDYNAMICSISPRESENTEDIN6FIG1THEBOILERANDSUPERHEATERSTEAMGENERATIONPROCESSFUZZYLOGICISCAPABLEOFINCORPORATINGHUMANEXPERIENCESVIATHEFUZZYRULESNEVERTHELESS,THEDESIGNOFFUZZYLOGICCONTROLLERSISSOMEHOWTIMECONSUMING,ASTHEFUZZYRULESAREOFTENOBTAINEDBYTRIALSANDERRORSINCONTRAST,NEURALNETWORKSNOTONLYHAVETHEABILITYTOAPPROXIMATENONLINEARFUNCTIONSWITHARBITRARYACCURACY,THEYCANALSOBETRAINEDFROMEXPERIMENTALDATATHENEUROFUZZYNETWORKSNFNSDEVELOPEDRECENTLYHAVETHEADVANTAGESOFMODELTRANSPARENCYOFFUZZYLOGIC,ANDLEARNINGCAPABILITYOFNEURALNETWORKS7THENFNSHAVEBEENUSEDTODEVELOPSELFRECEIVED14OCTOBER2005REVISED14OCTOBER2006THISWORKWASSUPPORTEDBYTHENATURALSCIENCEFOUNDATIONOFBEIJINGNO4062030,NATIONALNATURALSCIENCEFOUNDATIONOFCHINANO50576022,69804003,SCIENTIFICRESEARCHCOMMONPROGRAMOFBEIJINGMUNICIPALCOMMISSIONOFEDUCATIONKM200611232007XLIUETAL/JOURNALOFCONTROLTHEORYANDAPPLICATIONS20075183–88853NEUROFUZZYNETWORKGENERALIZEDPREDICTIVECONTROLTHEGPCISOBTAINEDBYMINIMIZINGTHEFOLLOWINGCOSTFUNCTION10,JEN?JDQJ?YTJ?YRTJ2M?J1ΛJΔUTJ?12,7WHEREQJANDΛJARERESPECTIVELYTHEWEIGHTINGFACTORSFORTHEPREDICTIONERRORANDTHECONTROL,YRTJISTHEJTHSTEPAHEADREFERENCETRAJECTORY,DISTHEMINIMUMCOSTINGHORIZON,NANDMARERESPECTIVELYTHEMAXIMUMCOSTINGHORIZONFORTHEPREDICTIONERRORANDTHECONTROLTHECONTROLCOMPUTEDFROMTHENFGPCISTHEWEIGHTEDSUMOFTHECONTROLOBTAINEDFROMPLOCALGPCCONTROLLERSΔUTP?I1ΑIΔUIT,8WHEREΔUITISTHECONTROLINTHEITHREGION,ΑIXISDEFINEDPREVIOUSLYIN4NOTETHATTHEWEIGHTSINTHENFGPCAREIDENTICALTOTHATINTHENFNTHATMODELSTHEPROCESSSINCESWITCHINGBETWEENLOCALGPCCONTROLLERSINTHENFGPCINVOLVESFUZZYLOGICS,ITCANBEINTERPRETEDNOTONLYASAFUZZYCONTROLLER,BUTALSOASAFUZZYSUPERVISORTHECONTROLCANBESMOOTHIFTHEWEIGHTSΑIXARESUITABLYSELECTEDFROMTHENFN6ANDTHECONTROL8,JGIVENBY7CANBEREWRITTENASJEN?JDQJP?I1ΑI?YITJ?YRTJ2M?J1ΛJP?I1ΑIΔUITJ?129THECOSTFUNCTIONISSIMPLIFIEDFIRSTUSINGTHECAUCHYINEQUALITYSINCEP?I1ΑI?YITJ?YRTJ2?PP?I1ΑI?YITJ?YRTJ2,HENCEP?I1ΑIΔUITJ?12?PP?I1ΑIΔUITJ?1210EQUATION10IMPLIESTHATTHESUMOFTHEWEIGHTEDSQUAREDERRORSCANBEANUPPERBOUNDOFTHECOSTFUNCTIONJREWRITING9GIVESEN?JDP?I1QJΑI?YTJ?YRTJ2M?J1P?I1ΛJΑIΔUITJ?12EP?I1ΑI2N?JDQJ?YITJ?YRTJ2P?I1ΑI2M?J1ΛJΔUITJ?12P?I1ΑI2JI,11WHEREJIEN?JDQJ?YITJ?YRTJ2M?J1ΛJΔUITJ?1212EQUATION11SHOWSTHATMINIMIZINGJIISESSENTIALLYTHESAMEASTHATOFMINIMIZINGJFROM12,ASETOFLOCALGENERALIZEDPREDICTIVECONTROLLERSISOBTAINED,WHICHFORMSPARTOFTHENFGPCTHELOCALGPC10ISGIVENBY,ΔUITGTIQIGIΛI?1GTIQIYRT1?FIΔUIT?1?SIZ?1YIT,13WHEREYRT1?YRT1,?YRT2,,?YRTNT,ΔUITΔUIT,ΔUIT1,,ΔUITM?1T,ΔUIT?1ΔUIT?NB,ΔUIT?NB1,,ΔUIT?1T,SIZ?1SI1Z?1,SI2Z?1,,SINZ?1TSIZ?1ANDRIZ?1SATISFYTHEDIOPHANTINEEQUATION1ˉAIZ?1RIJZ?1Z?JSIJZ?1,14ANDGIJZ?1BIZ?1RIJZ?1GIJ,0GIJ,1Z?1GIJ,NBJ?1Z?NBJ?1,15AQIDIAGQI1,QI2,,QIN,15BΛIDIAGΛI1,ΛI2,,ΛIM,15CGTI???????GI1,0GI2,1GIN,N?1GI1,0GIN?1,N?20GIN?M1,N?M???????,15DFI???????GI1,NBGI1,NB?1GI1,2GI1,1GI2,NB1GI2,NBGI2,3GI2,2GIN,NBN?1GIN,NBN?2GIN,N1GIN,N???????15ETHEOPTIMIZEDMSTEPSAHEADCONTROLISCOMPUTED,ANDONLYTHEFIRSTSTEPAHEADCONTROLISIMPLEMENTED,USINGARECEDINGHORIZONPRINCIPLE10,GIVINGΔUITDTI1YRT1?FIΔUIT?1?SIZ?1YIT,16

簡介：SAMPLEDDATAMODELPREDICTIVECONTROLFORNONLINEARTIMEVARYINGSYSTEMSSTABILITYANDROBUSTNESS?FERNANDOACCFONTES1,LALOMAGNI2,AND′EVAGYURKOVICS31OFFICINAMATHEMATICA,DEPARTAMENTODEMATEM′ATICAPARAACI?ENCIAETECNOLOGIA,UNIVERSIDADEDOMINHO,4800058GUIMAR?AES,PORTUGALFFONTESMCTUMINHOPT2DIPARTIMENTODIINFORMATICAESISTIMISTICA,UNIVERSITADEGLISTUDIDIPAVIA,VIAFERRATA1,27100PAVIA,ITALYLALOMAGNIUNIPVIT3BUDAPESTUNIVERSITYOFTECHNOLOGYANDECONOMICS,INSTITUTEOFMATHEMATICS,BUDAPESTH1521,HUNGARYGYEMATHBMEHUSUMMARYWEDESCRIBEHEREASAMPLEDDATAMODELPREDICTIVECONTROLFRAMEWORKTHATUSESCONTINUOUSTIMEMODELSBUTTHESAMPLINGOFTHEACTUALSTATEOFTHEPLANTASWELLASTHECOMPUTATIONOFTHECONTROLLAWS,ARECARRIEDOUTATDISCRETEINSTANTSOFTIMETHISFRAMEWORKCANADDRESSAVERYLARGECLASSOFSYSTEMS,NONLINEAR,TIMEVARYING,ANDNONHOLONOMICASINMANYOTHERSSAMPLEDDATAMODELPREDICTIVECONTROLSCHEMES,BARBALAT’SLEMMAHASANIMPORTANTROLEINTHEPROOFOFNOMINALSTABILITYRESULTSITISARGUEDTHATTHEGENERALIZATIONOFBARBALAT’SLEMMA,DESCRIBEDHERE,CANHAVEALSOASIMILARROLEINTHEPROOFOFROBUSTSTABILITYRESULTS,ALLOWINGALSOTOADDRESSAVERYGENERALCLASSOFNONLINEAR,TIMEVARYING,NONHOLONOMICSYSTEMS,SUBJECTTODISTURBANCESTHEPOSSIBILITYOFTHEFRAMEWORKTOACCOMMODATEDISCONTINUOUSFEEDBACKSISESSENTIALTOACHIEVEBOTHNOMINALSTABILITYANDROBUSTSTABILITYFORSUCHGENERALCLASSESOFSYSTEMS1INTRODUCTIONMANYMODELPREDICTIVECONTROLMPCSCHEMESDESCRIBEDINTHELITERATUREUSECONTINUOUSTIMEMODELSANDSAMPLETHESTATEOFTHEPLANTATDISCRETEINSTANTSOFTIMESEEEG3,7,9,13ANDALSO6THEREAREMANYADVANTAGESINCONSIDERINGACONTINUOUSTIMEMODELFORTHEPLANTNEVERTHELESS,ANYIMPLEMENTABLEMPCSCHEMECANONLYMEASURETHESTATEANDSOLVEANOPTIMIZATIONPROBLEMATDISCRETEINSTANTSOFTIMEINALLTHEREFERENCESCITEDABOVE,BARBALAT’SLEMMA,ORAMODIFICATIONOFIT,ISUSEDASANIMPORTANTSTEPTOPROVESTABILITYOFTHEMPCSCHEMESBARBALAT’S?THEFINANCIALSUPPORTFROMMURSTPROJECT“NEWTECHNIQUESFORTHEIDENTIFICATIONANDADAPTIVECONTROLOFINDUSTRIALSYSTEMS”,FROMFCTPROJECTPOCTI/MAT/61842/2004,ANDFROMTHEHUNGARIANNATIONALSCIENCEFOUNDATIONFORSCIENTIFICRESEARCHGRANTNOT037491ISGRATEFULLYACKNOWLEDGEDRFINDEISENETALEDSASSESSMENTANDFUTUREDIRECTIONS,LNCIS358,PP115–129,2007SPRINGERLINKCOMC?SPRINGERVERLAGBERLINHEIDELBERG2007SAMPLEDDATAMPCFORNONLINEARTIMEVARYINGSYSTEMS117ATTIMET0,AGIVENFUNCTIONFIRIRNIRM→IRN,ANDASETU?IRMOFPOSSIBLECONTROLVALUESWEASSUMETHISSYSTEMTOBEASYMPTOTICALLYCONTROLLABLEONX0ANDTHATFORALLT≥0FT,0,00WEFURTHERASSUMETHATTHEFUNCTIONFISCONTINUOUSANDLOCALLYLIPSCHITZWITHRESPECTTOTHESECONDARGUMENTTHECONSTRUCTIONOFTHEFEEDBACKLAWISACCOMPLISHEDBYUSINGASAMPLEDDATAMPCSTRATEGYCONSIDERASEQUENCEOFSAMPLINGINSTANTSΠ{TI}I≥0WITHACONSTANTINTERSAMPLINGTIMEΔ0SUCHTHATTI1TIΔFORALLI≥0CONSIDERALSOTHECONTROLHORIZONANDPREDICTIVEHORIZON,TCANDTP,WITHTP≥TCΔ,ANDANAUXILIARYCONTROLLAWKAUXIRIRN→IRMTHEFEEDBACKCONTROLISOBTAINEDBYREPEATEDLYSOLVINGONLINEOPENLOOPOPTIMALCONTROLPROBLEMSPTI,XTI,TC,TPATEACHSAMPLINGINSTANTTI∈Π,EVERYTIMEUSINGTHECURRENTMEASUREOFTHESTATEOFTHEPLANTXTIPT,XT,TC,TPMINIMIZETTP?TLS,XS,USDSWTTP,XTTP,2SUBJECTTO˙XSFS,XS,USAES∈T,TTP,3XTXT,XS∈XFORALLS∈T,TTP,US∈UAES∈T,TTC,USKAUXS,XSAES∈TTC,TTP,XTTP∈S4NOTETHATINTHEINTERVALTTC,TTPTHECONTROLVALUEISSELECTEDFROMASINGLETONANDTHEREFORETHEOPTIMIZATIONDECISIONSAREALLCARRIEDOUTINTHEINTERVALT,TTCWITHTHEEXPECTEDBENEFITSINTHECOMPUTATIONALTIMETHENOTATIONADOPTEDHEREISASFOLLOWSTHEVARIABLETREPRESENTSREALTIMEWHILEWERESERVESTODENOTETHETIMEVARIABLEUSEDINTHEPREDICTIONMODELTHEVECTORXTDENOTESTHEACTUALSTATEOFTHEPLANTMEASUREDATTIMETTHEPROCESSX,UISAPAIRTRAJECTORY/CONTROLOBTAINEDFROMTHEMODELOFTHESYSTEMTHETRAJECTORYISSOMETIMESDENOTEDASS?→XST,XT,UWHENWEWANTTOMAKEEXPLICITTHEDEPENDENCEONTHEINITIALTIME,INITIALSTATE,ANDCONTROLFUNCTIONTHEPAIRˉX,ˉUDENOTESOUROPTIMALSOLUTIONTOANOPENLOOPOPTIMALCONTROLPROBLEMTHEPROCESSX?,U?ISTHECLOSEDLOOPTRAJECTORYANDCONTROLRESULTINGFROMTHEMPCSTRATEGYWECALLDESIGNPARAMETERSTHEVARIABLESPRESENTINTHEOPENLOOPOPTIMALCONTROLPROBLEMTHATARENOTFROMTHESYSTEMMODELIEVARIABLESWEAREABLETOCHOOSETHESECOMPRISETHECONTROLHORIZONTC,THEPREDICTIONHORIZONTP,THERUNNINGCOSTANDTERMINALCOSTSFUNCTIONSLANDW,THEAUXILIARYCONTROLLAWKAUX,ANDTHETERMINALCONSTRAINTSETS?IRN

簡介：DOI101007/S0017000423288ORIGINALARTICLEINTJADVMANUFTECHNOL20062861–66FANGJUNGSHIOUCHAOCHANGACHENWENTULIAUTOMATEDSURFACEFINISHINGOFPLASTICINJECTIONMOLDSTEELWITHSPHERICALGRINDINGANDBALLBURNISHINGPROCESSESRECEIVED30MARCH2004/ACCEPTED5JULY2004/PUBLISHEDONLINE30MARCH2005?SPRINGERVERLAGLONDONLIMITED2005ABSTRACTTHISSTUDYINVESTIGATESTHEPOSSIBILITIESOFAUTOMATEDSPHERICALGRINDINGANDBALLBURNISHINGSURFACEFINISHINGPROCESSESINAFREEFORMSURFACEPLASTICINJECTIONMOLDSTEELPDS5ONACNCMACHININGCENTERTHEDESIGNANDMANUFACTUREOFAGRINDINGTOOLHOLDERHASBEENACCOMPLISHEDINTHISSTUDYTHEOPTIMALSURFACEGRINDINGPARAMETERSWEREDETERMINEDUSINGTAGUCHI’SORTHOGONALARRAYMETHODFORPLASTICINJECTIONMOLDINGSTEELPDS5ONAMACHININGCENTERTHEOPTIMALSURFACEGRINDINGPARAMETERSFORTHEPLASTICINJECTIONMOLDSTEELPDS5WERETHECOMBINATIONOFANABRASIVEMATERIALOFPAAL2O3,AGRINDINGSPEEDOF18000RPM,AGRINDINGDEPTHOF20ΜM,ANDAFEEDOF50MM/MINTHESURFACEROUGHNESSRAOFTHESPECIMENCANBEIMPROVEDFROMABOUT160ΜMTO035ΜMBYUSINGTHEOPTIMALPARAMETERSFORSURFACEGRINDINGSURFACEROUGHNESSRACANBEFURTHERIMPROVEDFROMABOUT0343ΜMTO006ΜMBYUSINGTHEBALLBURNISHINGPROCESSWITHTHEOPTIMALBURNISHINGPARAMETERSAPPLYINGTHEOPTIMALSURFACEGRINDINGANDBURNISHINGPARAMETERSSEQUENTIALLYTOAFINEMILLEDFREEFORMSURFACEMOLDINSERT,THESURFACEROUGHNESSRAOFFREEFORMSURFACEREGIONONTHETESTEDPARTCANBEIMPROVEDFROMABOUT215ΜMTO007ΜMKEYWORDSAUTOMATEDSURFACEFINISHINGBALLBURNISHINGPROCESSGRINDINGPROCESSSURFACEROUGHNESSTAGUCHI’SMETHOD1INTRODUCTIONPLASTICSAREIMPORTANTENGINEERINGMATERIALSDUETOTHEIRSPECIFICCHARACTERISTICS,SUCHASCORROSIONRESISTANCE,RESISTANCETOCHEMICALS,LOWDENSITY,ANDEASEOFMANUFACTURE,ANDHAVEINCREASINGLYFJSHIOUUCCACHENWTLIDEPARTMENTOFMECHANICALENGINEERING,NATIONALTAIWANUNIVERSITYOFSCIENCEANDTECHNOLOGY,NO43,SECTION4,KEELUNGROAD,106TAIPEI,TAIWANROCEMAILSHIOUMAILNTUSTEDUTWTEL886227376543FAX886227376460REPLACEDMETALLICCOMPONENTSININDUSTRIALAPPLICATIONSINJECTIONMOLDINGISONEOFTHEIMPORTANTFORMINGPROCESSESFORPLASTICPRODUCTSTHESURFACEFINISHQUALITYOFTHEPLASTICINJECTIONMOLDISANESSENTIALREQUIREMENTDUETOITSDIRECTEFFECTSONTHEAPPEARANCEOFTHEPLASTICPRODUCTFINISHINGPROCESSESSUCHASGRINDING,POLISHINGANDLAPPINGARECOMMONLYUSEDTOIMPROVETHESURFACEFINISHTHEMOUNTEDGRINDINGTOOLSWHEELSHAVEBEENWIDELYUSEDINCONVENTIONALMOLDANDDIEFINISHINGINDUSTRIESTHEGEOMETRICMODELOFMOUNTEDGRINDINGTOOLSFORAUTOMATEDSURFACEFINISHINGPROCESSESWASINTRODUCEDIN1AFINISHINGPROCESSMODELOFSPHERICALGRINDINGTOOLSFORAUTOMATEDSURFACEFINISHINGSYSTEMSWASDEVELOPEDIN2GRINDINGSPEED,DEPTHOFCUT,FEEDRATE,ANDWHEELPROPERTIESSUCHASABRASIVEMATERIALANDABRASIVEGRAINSIZE,ARETHEDOMINANTPARAMETERSFORTHESPHERICALGRINDINGPROCESS,ASSHOWNINFIG1THEOPTIMALSPHERICALGRINDINGPARAMETERSFORTHEINJECTIONMOLDSTEELHAVENOTYETBEENINVESTIGATEDBASEDINTHELITERATUREINRECENTYEARS,SOMERESEARCHHASBEENCARRIEDOUTINDETERMININGTHEOPTIMALPARAMETERSOFTHEBALLBURNISHINGPROCESSFIG2FORINSTANCE,ITHASBEENFOUNDTHATPLASTICDEFORMATIONONTHEWORKPIECESURFACECANBEREDUCEDBYUSINGATUNGSTENCARBIDEBALLORAROLLER,THUSIMPROVINGTHESURFACEROUGHNESS,SURFACEHARDNESS,ANDFATIGUERESISTANCE3–6THEBURNISHINGPROCESSISACCOMPLISHEDBYMACHININGCENTERS3,4ANDLATHES5,6THEMAINBURNISHINGPARAMETERSHAVINGSIGNIFICANTEFFECTSONTHESURFACEROUGHNESSAREBALLORROLLERMATERIAL,BURNISHINGFORCE,FEEDRATE,BURNISHINGSPEED,LUBRICATION,ANDNUMBEROFBURNISHINGPASSES,AMONGOTHERS3THEOPTIMALSURFACEBURNISHINGPARAMETERSFORTHEPLASTICINJECTIONMOLDSTEELPDS5WEREACOMBINATIONOFGREASELUBRICANT,THETUNGSTENCARBIDEBALL,ABURNISHINGSPEEDOF200MM/MIN,ABURNISHINGFORCEOF300N,ANDAFEEDOF40ΜM7THEDEPTHOFPENETRATIONOFTHEBURNISHEDSURFACEUSINGTHEOPTIMALBALLBURNISHINGPARAMETERSWASABOUT25MICRONSTHEIMPROVEMENTOFTHESURFACEROUGHNESSTHROUGHBURNISHINGPROCESSGENERALLYRANGEDBETWEEN40AND903–7THEAIMOFTHISSTUDYWASTODEVELOPSPHERICALGRINDINGANDBALLBURNISHINGSURFACEFINISHPROCESSESOFAFREEFORMSURFACE63FIG4SCHEMATICILLUSTRATIONOFTHESPHERICALGRINDINGTOOLANDITSADJUSTMENTDEVICE3PLANNINGOFTHEMATRIXEXPERIMENT31CONFIGURATIONOFTAGUCHI’SORTHOGONALARRAYTHEEFFECTSOFSEVERALPARAMETERSCANBEDETERMINEDEFFICIENTLYBYCONDUCTINGMATRIXEXPERIMENTSUSINGTAGUCHI’SORTHOGONALARRAY8TOMATCHTHEAFOREMENTIONEDSPHERICALGRINDINGPARAMETERS,THEABRASIVEMATERIALOFTHEGRINDERBALLWITHTHEDIAMETEROF10MM,THEFEEDRATE,THEDEPTHOFGRINDING,ANDTHEREVOLUTIONOFTHEELECTRICGRINDERWERESELECTEDASTHEFOUREXPERIMENTALFACTORSPARAMETERSANDDESIGNATEDASFACTORATODSEETABLE1INTHISRESEARCHTHREELEVELSSETTINGSFOREACHFACTORWERECONFIGUREDTOCOVERTHERANGEOFINTEREST,ANDWEREIDENTIFIG5APHOTOOFTHESPHERICALGRINDINGTOOLBPHOTOOFTHEBALLBURNISHINGTOOLTABLE1THEEXPERIMENTALFACTORSANDTHEIRLEVELSFACTORLEVEL123AABRASIVEMATERIALSICAL2O3,WAAL2O3,PABFEEDMM/MIN50100200CDEPTHOFGRINDINGΜM205080DREVOLUTIONRPM120001800024000FIEDBYTHEDIGITS1,2,AND3THREETYPESOFABRASIVEMATERIALS,NAMELYSILICONCARBIDESIC,WHITEALUMINUMOXIDEAL2O3,WA,ANDPINKALUMINUMOXIDEAL2O3,PA,WERESELECTEDANDSTUDIEDTHREENUMERICALVALUESOFEACHFACTORWEREDETERMINEDBASEDONTHEPRESTUDYRESULTSTHEL18ORTHOGONALARRAYWASSELECTEDTOCONDUCTTHEMATRIXEXPERIMENTFORFOUR3LEVELFACTORSOFTHESPHERICALGRINDINGPROCESS32DEFINITIONOFTHEDATAANALYSISENGINEERINGDESIGNPROBLEMSCANBEDIVIDEDINTOSMALLERTHEBETTERTYPES,NOMINALTHEBESTTYPES,LARGERTHEBETTERTYPES,SIGNEDTARGETTYPES,AMONGOTHERS8THESIGNALTONOISES/NRATIOISUSEDASTHEOBJECTIVEFUNCTIONFOROPTIMIZINGAPRODUCTORPROCESSDESIGNTHESURFACEROUGHNESSVALUEOFTHEGROUNDSURFACEVIAANADEQUATECOMBINATIONOFGRINDINGPARAMETERSSHOULDBESMALLERTHANTHATOFTHEORIGINALSURFACECONSEQUENTLY,THESPHERICALGRINDINGPROCESSISANEXAMPLEOFASMALLERTHEBETTERTYPEPROBLEMTHES/NRATIO,Η,ISDEFINEDBYTHEFOLLOWINGEQUATION8Η?10LOG10MEANSQUAREQUALITYCHARACTERISTIC?10LOG10?1NN?I1Y2I?1WHEREYIOBSERVATIONSOFTHEQUALITYCHARACTERISTICUNDERDIFFERENTNOISECONDITIONSNNUMBEROFEXPERIMENTAFTERTHES/NRATIOFROMTHEEXPERIMENTALDATAOFEACHL18ORTHOGONALARRAYISCALCULATED,THEMAINEFFECTOFEACHFACTORWASDETERMINEDBYUSINGANANALYSISOFVARIANCEANOVATECHNIQUEANDANFRATIOTEST8THEOPTIMIZATIONSTRATEGYOFTHE

簡介：中文中文65006500字出處IEEETRANSACTIONSONCONSUMERELECTRONICS,1094VOL50,NO4,NOVEMBER2004基于區(qū)域控制網(wǎng)絡(luò)CAN的智能家居自動化火災(zāi)報警系統(tǒng)KYUNGCHANGLEE,HONGHEELEE摘要本文提出一個應(yīng)用區(qū)域控制網(wǎng)絡(luò)CAN的火災(zāi)報警系統(tǒng)并評估其應(yīng)用于智能家庭自動化控制的可能性。通常，傳統(tǒng)的火災(zāi)報警系統(tǒng)有一些不足，例如由于其使用420MA的模擬電流信號，噪聲對其干擾很大。因此，本文為替代原有系統(tǒng)，提出了一個基于CAN的火災(zāi)報警系統(tǒng)，闡述了CAN通信網(wǎng)絡(luò)的設(shè)計方法并進行試驗評估該系統(tǒng)的性能。這個網(wǎng)絡(luò)有以下幾個優(yōu)點，如比其他底層BACNET如以太網(wǎng)、ARCNET有更低的成本且更容易實現(xiàn)。因此，如果CAN被選為底層BACNET，家庭自動化系統(tǒng)將會更有效。關(guān)鍵詞網(wǎng)絡(luò)型火災(zāi)報警系統(tǒng)；區(qū)域控制網(wǎng)絡(luò)CAN家庭自動化系統(tǒng)；家庭網(wǎng)絡(luò)系統(tǒng)；智能建筑1引言當(dāng)前，建筑的智能化為人們帶來更多的方便與安全12。因此，家庭網(wǎng)絡(luò)自動化系統(tǒng)的要求隨著智能家居要求的增長而日益增長3。為了滿足使用者的需求、家電如冰箱和微波爐、多媒體設(shè)備如電視和音響系統(tǒng)和網(wǎng)絡(luò)設(shè)備如電腦已被包括在智能建筑，如圖1。在智能家居，我們可以在房內(nèi)或是戶外用一個手機或PDA監(jiān)控連接到家庭網(wǎng)絡(luò)的電器。為了實現(xiàn)家庭網(wǎng)絡(luò)系統(tǒng)，一些標準組織、企業(yè)正在開發(fā)ECHONET,KONNEX,LNCP和LONWORKS等網(wǎng)絡(luò)標準4。圖1家庭網(wǎng)絡(luò)系統(tǒng)原理圖另外，為了提高人們生活的舒適與安全正在完善如強電控制、照明、防盜、火災(zāi)報警等家庭自動化系統(tǒng)。通常，在傳統(tǒng)的家庭自動化系統(tǒng)中，開關(guān)、閥門或者火災(zāi)探測器都直接與空調(diào)設(shè)備或火災(zāi)報警系統(tǒng)相連。傳統(tǒng)火災(zāi)告警系統(tǒng)采用420MA電流的模擬傳輸方式，當(dāng)從火災(zāi)探測器接受的電流信號超過閾值，判定發(fā)生火災(zāi)。因此，該系統(tǒng)存在一些不足，它容易受到包括尖脈沖等不同形式的干擾，同時它不能判斷實際的燃火點。為了解決這些問題，業(yè)界已經(jīng)開始研究用數(shù)字、無線傳輸INTERNET移動電腦手機、PDA等家庭網(wǎng)關(guān)家庭應(yīng)用控制網(wǎng)絡(luò)骨干網(wǎng)信息網(wǎng)絡(luò)多媒體網(wǎng)絡(luò)多媒體設(shè)備信息設(shè)備強電控制照明控制取暖控制家庭網(wǎng)絡(luò)系統(tǒng)家庭自動化系統(tǒng)過這種連接方式，由于每一個火災(zāi)探測器都有屬于自己的唯一地址，接收器就可以識別是哪個探測器進行告警。此外，由于接收器定期檢測各個火災(zāi)探測器的狀態(tài)，它可以發(fā)現(xiàn)諸如探測器故障或是傳輸總線開路等系統(tǒng)故障。另外，因為各個探測器將煙霧與熱度的定量數(shù)據(jù)發(fā)送給接收器，所以錯誤警報要少于傳統(tǒng)火災(zāi)報警系統(tǒng)。在同一區(qū)域安裝多個火災(zāi)探測器，接收器可以到各位置直觀的煙霧、熱度數(shù)據(jù)，因此本系統(tǒng)可以直接應(yīng)用于智能火災(zāi)報警系統(tǒng)并能使用推理算法。同時由于探測信號可以進行計算比較，系統(tǒng)的風(fēng)險評估能力得到提升12。圖3網(wǎng)絡(luò)型火災(zāi)報警系統(tǒng)的網(wǎng)絡(luò)結(jié)構(gòu)此外，由于PC的安裝維護比傳統(tǒng)的專用接收機更為方便，我們可以很容易地創(chuàng)建一個用戶界面，將人機界面應(yīng)用于PC，將本系統(tǒng)合成到家庭網(wǎng)絡(luò)系統(tǒng)中，因此網(wǎng)絡(luò)火災(zāi)報警系統(tǒng)如果使用PC作為接收機會更為方便。當(dāng)前，已有BACNET、LONWORKS、BLUETOOTH等多種網(wǎng)絡(luò)協(xié)議將網(wǎng)絡(luò)型火災(zāi)告警系統(tǒng)應(yīng)用在智能建筑項目用56。3網(wǎng)絡(luò)型火災(zāi)報警系統(tǒng)的設(shè)計31CAN協(xié)議概述CAN20B是專門為連接傳感器、驅(qū)動器、汽車中的電子控制單元ECU而編寫的網(wǎng)絡(luò)協(xié)議。它的支持通信速率為5KBPS1MBPS，可以應(yīng)用于信息共享與實時控制領(lǐng)域。它可以選擇總線型或者星型網(wǎng)絡(luò)拓撲結(jié)構(gòu)。CAN20B具有以下性質(zhì)分布式總線存取控制。這意味著每一個設(shè)備都有相同的總線使用權(quán)限。競爭非破壞性總線讀取。雖然設(shè)配是通過競爭方式控制總線，但不會因競爭破壞報文。根據(jù)內(nèi)容確定地址。每條報文根據(jù)自身內(nèi)容確定唯一的標識。循環(huán)冗余校驗錯誤檢測和錯誤禁閉來阻止任何不利影響的一個網(wǎng)絡(luò)元件失效。設(shè)備不管網(wǎng)絡(luò)是否空閑都可以傳輸信息。當(dāng)網(wǎng)絡(luò)繁忙時，在正在傳輸?shù)男畔瓿汕?，將要發(fā)送的包會一直等待。電信號在網(wǎng)絡(luò)中傳輸?shù)乃俣仁怯邢薜?，很有可能多個設(shè)備在很短的時間段內(nèi)都要開始傳輸信息。這種情況被稱為信息沖突，協(xié)議通過比較報文包含的標識來解決這一難題。標識的值最小的報文贏得網(wǎng)絡(luò)使用權(quán)，其他的設(shè)備必須立即停止傳輸。因為標識在數(shù)據(jù)包的首部，一起傳輸?shù)碾娦盘枴?”將把電信號“1”改寫，所以標識值最小的數(shù)據(jù)包將不被破壞的完成傳輸。其他的設(shè)備將在當(dāng)前傳輸結(jié)束后，繼續(xù)嘗試傳輸自己的數(shù)據(jù)。圖4展現(xiàn)這個仲裁的過程。響鈴指示燈火災(zāi)探測器驅(qū)動器PC端接收器網(wǎng)絡(luò)總線BDDA

簡介：SHIFTDYNAMICSANDCONTROLOFDUALCLUTCHTRANSMISSIONSMANISHKULKARNI,TAEHYUNSHIM,YIZHANGDEPARTMENTOFMECHANICALENGINEERING,UNIVERSITYOFMICHIGANDEARBORN,DEARBORNMI48128,UNITEDSTATESRECEIVED4OCTOBER2005ACCEPTED1MARCH2006AVAILABLEONLINE18MAY2006ABSTRACTSHIFTSINADUALCLUTCHTRANSMISSIONDCTAREREALIZEDBYTORQUETRANSFERFROMONECLUTCHTOANOTHERWITHOUTTRACTIONINTERRUPTIONDUETOTHECONTROLLEDSLIPPAGEOFTHECLUTCHESTHETIMINGOFENGAGEMENTANDDISENGAGEMENTOFTHETWOCLUTCHESISCRITICALFORACHIEVINGASMOOTHSHIFTWITHOUTENGINEFLAREANDCLUTCHTIEUPTHISPAPERPRESENTSANANALYTICALMODELFORTHESIMULATION,ANALYSISANDCONTROLOFSHIFTDYNAMICSFORDCTVEHICLESADYNAMICMODELANDTHECONTROLLOGICFORTHEINTEGRATEDVEHICLEHAVEBEENDEVELOPEDUSINGMATLAB/SIMULINKASTHESIMULATIONPLATFORMTHEMODELHASBEENUSEDTOSTUDYTHEVARIATIONINOUTPUTTORQUEINRESPONSETODIFFERENTCLUTCHPRESSUREPROFILESDURINGSHIFTSOPTIMIZEDCLUTCHPRESSUREPROFILESHAVEBEENCREATEDFORTHEBESTPOSSIBLESHIFTQUALITYBASEDONMODELSIMULATIONASANUMERICALEXAMPLE,THEMODELISUSEDFORADCTVEHICLETOSIMULATETHEWIDEOPENTHROTTLEPERFORMANCEVEHICLELAUNCHANDSHIFTPROCESSAREBOTHSIMULATEDTOASSESSTRANSMISSIONSHIFTQUALITYANDVALIDATETHEEFFECTIVENESSOFTHESHIFTCONTROL?2006ELSEVIERLTDALLRIGHTSRESERVEDKEYWORDSDUALCLUTCHTRANSMISSIONAUTOMATICTRANSMISSIONS1INTRODUCTIONTHEREHASBEENACLEARTRENDINTHEAUTOMOTIVEINDUSTRYINRECENTYEARSTOWARDSINCREASEDRIDECOMFORTANDFUELEFFICIENCYASTHEPOWERTRANSMISSIONUNIT,TRANSMISSIONSPLAYANIMPORTANTROLEINVEHICLEPERFORMANCEANDFUELECONOMYTHEREARECURRENTLYSEVERALTYPESOFTRANSMISSIONSANDTHEASSOCIATEDTECHNOLOGIESTHATOFFERDIFFERENTPERFORMANCEPRIORITIESWHENFITINTOAVEHICLE1MANUALTRANSMISSIONSHAVEANOVERALLEFFICIENCYOF962,WHICHISTHEHIGHESTEFFICIENCYVALUEFORANYTYPEOFTRANSMISSIONCURRENTPRODUCTIONAUTOMATICSHAVEBEENIMPROVEDTOPROVIDEANEFFICIENCYOFNOTMORETHAN863BELTTYPECVT’SHAVEANOVERALLEFFICIENCYOF846,HOWEVER,THEMAJORADVANTAGEOFCVTISTHATITALLOWSTHEENGINETOOPERATEINTHEMOSTFUELEFFICIENTMANNER2AUTOMATEDMANUALTRANSMISSIONSHAVETHESAMEEFFICIENCYOFMANUALTRANSMISSIONSANDOFFEROPERATIONCONVENIENCESIMILARTOCONVENTIONALAUTOMATICTRANSMISSIONSTHEREEXISTTWOTECHNICALLYFEASIBLEDESIGNSFORAUTOMATEDLAYSHAFTGEARINGTRANSMISSIONSONEUSESASINGLECLUTCHANDISBASICALLYAMANUALTRANSMISSIONWITHANADDEDONCONTROLUNITTHATAUTOMATESTHECLUTCHANDSHIFTOPERATIONSINTHISDESIGN,THEREISANINTERRUPTIONOF0094114X/SEEFRONTMATTER?2006ELSEVIERLTDALLRIGHTSRESERVEDDOI101016/JMECHMACHTHEORY200603002CORRESPONDINGAUTHORTEL13135935539EMAILADDRESSANDINGUMICHEDUYZHANGMECHANISMANDMACHINETHEORY422007168–182WWWELSEVIERCOM/LOCATE/MECHMTMECHANISMANDMACHINETHEORYTHEENGINEOUTPUTTORQUEISINTERPOLATEDINTERMSOFTHETHROTTLEANGLEANDRPMFROMTHEENGINEMAPGEARSHAVENOBACKLASHALLTHEMECHANICALLOSSESAREMODELEDASAPARTOFTHEVEHICLEDRAGDELAYSDUETOHYDRAULICACTUATIONSYSTEMARENOTCONSIDEREDCLUTCHESAREMODELEDASCOULOMBFRICTIONELEMENTSTEMPERATUREEFFECTSOFTHEDRIVETRAINARENEGLECTEDCL2CL1SYN5RSYN6SYN13SYN244231INPUTSHAFTOUTPUT6R5FINALDRIVEPINION2FINALDRIVEPINION1INTERMEDIATESHAFT1INTERMEDIATESHAFT2SYN5RSYN6SYN13SYN24INPUTSHAFTOUTPUT6R5SYN5RSYN6SYN13SYN24INPUTSHAFTOUTPUT6R5FINALDRIVEPINION2FINALDRIVEPINION1INTERMEDIATESHAFT1INTERMEDIATESHAFT2FIG1DCTSTICKDIAGRAM4OUTPUTSHAFT4RCL1CL2131265INPUTSHAFTII/PIECL1312ENGINEKMCMIMEΩPI/ΩK1C1IHISI1SΩI2IMΩIMΩHΩK2C2I3AIOΩWΩFIG2DCTDYNAMICMODEL170MKULKARNIETAL/MECHANISMANDMACHINETHEORY422007168–182

簡介：STRUCTMULTIDISCOPTIM20,76–82?SPRINGERVERLAG2000OPTIMALDESIGNOFHYDRAULICSUPPORTMOBLAK,BHARLANDBBUTINARABSTRACTTHISPAPERDESCRIBESAPROCEDUREFOROPTIMALDETERMINATIONOFTWOGROUPSOFPARAMETERSOFAHYDRAULICSUPPORTEMPLOYEDINTHEMININGINDUSTRYTHEPROCEDUREISBASEDONMATHEMATICALPROGRAMMINGMETHODSINTHEFIRSTSTEP,THEOPTIMALVALUESOFSOMEPARAMETERSOFTHELEADINGFOURBARMECHANISMAREFOUNDINORDERTOENSURETHEDESIREDMOTIONOFTHESUPPORTWITHMINIMALTRANSVERSALDISPLACEMENTSINTHESECONDSTEP,MAXIMALTOLERANCESOFTHEOPTIMALVALUESOFTHELEADINGFOURBARMECHANISMARECALCULATED,SOTHERESPONSEOFHYDRAULICSUPPORTWILLBESATISFYINGKEYWORDSFOURBARMECHANISM,OPTIMALDESIGN,MATHEMATICALPROGRAMMING,APPROXIMATIONMETHOD,TOLERANCE1INTRODUCTIONTHEDESIGNERAIMSTOFINDTHEBESTDESIGNFORTHEMECHANICALSYSTEMCONSIDEREDPARTOFTHISEFFORTISTHEOPTIMALCHOICEOFSOMESELECTEDPARAMETERSOFASYSTEMMETHODSOFMATHEMATICALPROGRAMMINGCANBEUSED,IFASUITABLEMATHEMATICALMODELOFTHESYSTEMISMADEOFCOURSE,ITDEPENDSONTHETYPEOFTHESYSTEMWITHTHISFORMULATION,GOODCOMPUTERSUPPORTISASSUREDTOLOOKFOROPTIMALPARAMETERSOFTHESYSTEMTHEHYDRAULICSUPPORTFIG1DESCRIBEDBYHARL1998ISAPARTOFTHEMININGINDUSTRYEQUIPMENTINTHEMINEVELENJESLOVENIA,USEDFORPROTECTIONOFWORKINGENVIRONMENTINTHEGALLERYITCONSISTSOFTWOFOURBARRECEIVEDAPRIL13,1999MOBLAK1,BHARL2ANDBBUTINAR31FACULTYOFMECHANICALENGINEERING,SMETANOVA17,2000MARIBOR,SLOVENIAEMAILMAKSOBLAKUNIMBSI2MPPRAZVOJDOO,PTUJSKA184,2000MARIBOR,SLOVENIAEMAILBOSTJANHARLUNIMBSI3FACULTYOFCHEMISTRYANDCHEMICALENGINEERING,SMETANOVA17,2000MARIBOR,SLOVENIAEMAILBRANKOBUTINARUNIMBSIMECHANISMSFEDGANDAEDBASSHOWNINFIG2THEMECHANISMAEDBDEFINESTHEPATHOFCOUPLERPOINTCANDTHEMECHANISMFEDGISUSEDTODRIVETHESUPPORTBYAHYDRAULICACTUATORFIG1HYDRAULICSUPPORTITISREQUIREDTHATTHEMOTIONOFTHESUPPORT,MOREPRECISELY,THEMOTIONOFPOINTCINFIG2,ISVERTICALWITHMINIMALTRANSVERSALDISPLACEMENTSIFTHISISNOTTHECASE,THEHYDRAULICSUPPORTWILLNOTWORKPROPERLYBECAUSEITISSTRANDEDONREMOVALOFTHEEARTHMACHINEAPROTOTYPEOFTHEHYDRAULICSUPPORTWASTESTEDINALABORATORYGRM1992THESUPPORTEXHIBITEDLARGETRANSVERSALDISPLACEMENTS,WHICHWOULDREDUCEITSEMPLOYABILITYTHEREFORE,AREDESIGNWASNECESSARYTHEPROJECTSHOULDBEIMPROVEDWITHMINIMALCOSTIFPOS7821MATHEMATICALMODELTHEMATHEMATICALMODELOFTHESYSTEMWILLBEFORMULATEDINTHEFORMPROPOSEDBYHAUGANDARORA1979MINFU,V,9SUBJECTTOCONSTRAINTSGIU,V≤0,I1,2,,?,10ANDRESPONSEEQUATIONSHJU,V0,J1,2,,M11THEVECTORUU1UNTISCALLEDTHEVECTOROFDESIGNVARIABLES,VV1VMTISTHEVECTOROFRESPONSEVARIABLESANDFIN9ISTHEOBJECTIVEFUNCTIONTOPERFORMTHEOPTIMALDESIGNOFTHELEADINGFOURBARMECHANISMAEDB,THEVECTOROFDESIGNVARIABLESISDEFINEDASUA1A2A4T,12ANDTHEVECTOROFRESPONSEVARIABLESASVXYT13THEDIMENSIONSA3,A5,A6OFTHECORRESPONDINGLINKSAREKEPTFIXEDTHEOBJECTIVEFUNCTIONISDEFINEDASSOME“MEASUREOFDIFFERENCE”BETWEENTHETRAJECTORYLANDTHEDESIREDTRAJECTORYKASFU,VMAXG0Y?F0Y2,14WHEREXG0YISTHEEQUATIONOFTHECURVEKANDXF0YISTHEEQUATIONOFTHECURVELSUITABLELIMITATIONSFOROURSYSTEMWILLBECHOSENTHESYSTEMMUSTSATISFYTHEWELLKNOWNGRASSHOFFCONDITIONSA3A4?A1A2≤0,15A2A3?A1A4≤016INEQUALITIES15AND16EXPRESSTHEPROPERTYOFAFOURBARMECHANISM,WHERETHELINKSA2,A4MAYONLYOSCILLATETHECONDITIONU≤U≤U17PRESCRIBESTHELOWERANDUPPERBOUNDSOFTHEDESIGNVARIABLESTHEPROBLEM9–11ISNOTDIRECTLYSOLVABLEWITHTHEUSUALGRADIENTBASEDOPTIMIZATIONMETHODSTHISCOULDBECIRCUMVENTEDBYINTRODUCINGANARTIFICIALDESIGNVARIABLEUN1ASPROPOSEDBYHSIEHANDARORA1984THENEWFORMULATIONEXHIBITINGAMORECONVENIENTFORMMAYBEWRITTENASMINUN1,18SUBJECTTOGIU,V≤0,I1,2,,?,19FU,V?UN1≤0,20ANDRESPONSEEQUATIONSHJU,V0,J1,2,,M,21WHEREUU1UNUN1TANDVV1VMTANONLINEARPROGRAMMINGPROBLEMOFTHELEADINGFOURBARMECHANISMAEDBCANTHEREFOREBEDEFINEDASMINA7,22SUBJECTTOCONSTRAINTSA3A4?A1A2≤0,23A2A3?A1A4≤0,24A1≤A1≤A1,A2≤A2≤A2,A4≤A4≤A4,25G0Y?F0Y2?A7≤0,Y∈??Y,Y??,26ANDRESPONSEEQUATIONSX?A5COSΘ2Y?A5SINΘ2?A220,27X?A6COSΘΓ?A12Y?A6SINΘΓ2?A24028THISFORMULATIONENABLESTHEMINIMIZATIONOFTHEDIFFERENCEBETWEENTHETRANSVERSALDISPLACEMENTOFTHEPOINTCANDTHEPRESCRIBEDTRAJECTORYKTHERESULTISTHEOPTIMALVALUESOFTHEPARAMETERSA1,A2,A4