中国机械工程学报(英文版)论文模板

./下行页眉字号9磅,单倍行距,段后1.2磅字号8磅,段后回车换行2次首页上行页眉字号10.5磅,单倍行距,段后1.2磅DOI:10.3901/CJME.20下行页眉字号9磅,单倍行距,段后1.2磅字号8磅,段后回车换行2次首页上行页眉字号10.5磅,单倍行距,段后1.2磅论文题目字号论文题目字号14磅,行距固定值16磅,段后回车换行2次字号11磅字号11磅,段后8磅ZHANGJiafan1,3,*,FUHailun2,DONGYiming1,ZHANGYu1,YANGCanjun1,andCHENYing11StateKeyLaboratoryofFluidPowerTransmissionandControl,ZhejiangUniversity,Hangzhou310027,字号10磅,斜体,段后回车换行1次2ZhejiangProvinceInstituteofMetrology,字号10磅,斜体,段后回车换行1次3NationalDie&MoldCADEngineeringResearchCenter,ShanghaiJiaoTongUnivReceivedMarch1,2014;revisedJune17,2014;acceptedAugust17,2014字号字号8磅,段后回车换行1次字号9磅,段后回车换行1次Abstract:Magneticdrivepumphasgottengreatachievementandhasbeenwidelyusedinsomespecialfields.Currently,theresearchesonmagneticdrivepumphavefocusedonhydraulicdesign,bearing,axialforceinChina,andanewmagneticdrivepumpwithlowflowandhighheadhavebeendevelopedoverseas.However,lowefficiencyandlargesizearethecommondisadvantagesforthemagneticdrivepump.Inordertostudytheperformanceofhigh-speedmagneticdrivepump,FLUENTisusedtosimulatetheinnerflowfieldofmagneticdrivepumpswithdifferentrotatespeeds,andgetvelocityandpressuredistributionsofinnerflowfield.Accordingtoanalysisthechangesofvelocityandpressuretoensurethestableoperationofpumpandavoidcavitation.Basedontheanalysisofvelocityandpressure,thispaperpresentsthepumpefficiencyofmagneticdrivepumpswithdifferentrotatedspeedsbycalculatingthepowerlossinimpellerandvolute,hydraulicloss,volumetricloss,mechanicallossanddiscussingthedifferentreasonsofpowerlossbetweenthemagneticdrivepumpswithdifferentrotatedspeeds.Inaddition,themagneticdrivepumpsaretestedinaclosedtestingsystem.Pressuresensorsaresetininletandoutletofmagneticdrivepumpstogetthepressureandthehead,whilethepumpefficiencycanbegotbycalculatingthepowerlossbetweentheinputpowerandtheoutletpower.Theresultsofsimulationandtestaresimilar,whichshowsthatthemethodofsimulationisfeasible.Theproposedresearchprovidestheinstructionto字号9磅,段后回车换行1次Keywords:exoskeletonarm,teleoperation,pneumaticforce-feedback,hybridfuzzycontrol一级标题字号12磅一级标题字号12磅,段前空1行,段后回车换行1次字号9磅,段后回车换行2次.1Introduction**Correspondingauthor.E-mail:caffeezhang@SupportedbyNationalNaturalScienceFoundationofChina<GrantNo.50305035>,NationalHi-techResearchandDevelopmentProgramofChina<863Program,GrantNo.##>,BeijingMunicipalNaturalScienceFoundationofChina<<GrantNo.##>,andZhejiangProvincialNaturalScienceFoundationofChina<<GrantNo.##>©ChineseMechanicalEngineeringSocietyandSpringer-VerlagBerlinHeidelberg2014正文字号为10磅,除特别标注外,行距均为固定值13磅,行首缩进0.35cm；含有上下标或较复杂数学式的段落为多倍行距,设置值为1.1~1.2字号8磅,单倍行距菜单栏"格式"中的"段落",其中"缩进和间距"标签的复选框"如果定义了文档网络,则自动调整右缩进〔D"和"如果定义了文档网络,则对齐网格〔W"不要勾选Atfirstlookatmodernsociety,moreandmorerobotsandautomateddevicesarecomingintoourlifeandserveforhuman.Butonevenfurtherinspection,onecanfindthatmechatronicdevicesreplacehumansubordinatelyonlyatlowerlevels,essentiallyprovidingthe"grunt"toperformroutinetasks.Humancontrolisstillnecessaryatallhigherlevelsjustasthetermhumansupervisorycontrol<HSC>,whichiscoinedbySHERIDAN[1].Themodernmaster-slaveteleoperationsystemforthesafemanipulationofradioactivematerialsinacontaminatedareain1954ofGOERTZ,etal[2],wasthetypicalexampleofthisconcept.Hereafter,exoskeletonarmswithforce-feedbackhavebeenwidelydevelopedinthefieldsofrobotteleoperationandhapticinterfacetoenhancetheperformanceofthehumanoperator,alsointheexcitingapplicationsinsurgeryplanning,personneltraining,andphysicalrehabilitation.DUBEY,etal[3],developedamethodologytoincorporatesensorandmodelbasedcomputerassistanceintohumancontrolledteleoperationsystems.Intheirapproach,thehumanoperatorwasretainedatallphasesoftheoperation,andwasassistedbyadjustingsystemparameterswhichwerenotunderdirectcontrolbytheoperator,specifically,themappingofpositionsandvelocitiesbetweenthemasterandslaveandtheirimpedanceparameters.TheESAhumanarmexoskele-tonwasdevelopedtoenableforce-feedbacktele-manipulationontheexterioroftheinternationalspacestationwithredundantroboticarms[4].Inrecentwork[5–6],theneuromuscularsignalhasbeenusedtocontroltheexoskeletonarmandmanynewconceptswereappliedintherehabilitation[7–10].SeveralresearchersfromKoreaInstitutionofScienceandTechnology<KIST>introducedthepneumaticactuatorintotheexoskeletonanddesigned正文字号为10磅,除特别标注外,行距均为固定值13磅,行首缩进0.35cm；含有上下标或较复杂数学式的段落为多倍行距,设置值为1.1~1.2字号8磅,单倍行距菜单栏"格式"中的"段落",其中"缩进和间距"标签的复选框"如果定义了文档网络,则自动调整右缩进〔D"和"如果定义了文档网络,则对齐网格〔W"不要勾选字号6.5磅,单倍行距Inthisresearch,awearableexoskeletonarm,ZJUESA,basedonman-machinesystemisdesignedandahierarchicallydistributedteleoperationcontrolsystemisexplained.Thissystemincludesthreemainlevels:①supervisorgivingthecommandthroughtheexoskeletonarminsafezonewiththeoperatorinterface;②slave-robotworkinginhazardouszone;③datatransmissionbetweensupervisor-masterandmaster-slavethroughtheInternetorEthernet.Insection2,byusingtheorthogonalexperimentdesignmethod,thedesignfoundationofZJUESAanditsoptimaldesignarepresented.Theninsection3,wedescribeanovelhybridfuzzycontrolsystemfortheforcefeedbackonZJUESA.Consequently,theforcefeedbackcontrolsimulationsandexperimentresultsanalysisarepresentedinsection4字号6.5磅双码页面页眉字号双码页面页眉字号8磅,单倍行距,段后1.2磅2ConfigurationoftheExoskeletonArm

System图片及表格要求：图表类型分辨率灰度<150|>225彩色图片及表格要求：图表类型分辨率灰度<150|>225彩色<150|>225位图<600|>900图前段落,段后回车换行1次Fig.1.Configurationoftheexoskeletonarmsystem图题字号9磅,行距固定值11磅,段前0.3行,段后回车换行1次；图中字号8磅图题字号9磅,行距固定值11磅SinceZJUESAisdesignedbyfollowingthephysiologicalparametersofthehumanupper-limb,withsuchadevicethehumanoperatorcancontrolthemanipulatormorecomfortablyandintuitivelythanthesystemwiththejoystickorthekeyboardinput.3DesignoftheExoskeletonArm二级标题字号10磅二级标题字号10磅Whatwedesireisanarmexoskeletonwhichiscapableoffollowingmotionsofthehumanupper-limbaccuratelyandsupplyingthehumanupper-limbwithproperforcefeedbackifneeded.Inordertoachieveanidealcontrollingperformance,wehavetoexaminethestructureofthehuman二级标题字号10磅二级标题字号10磅3.1Anatomyofhumanupper-limb三级标题字号10磅,斜体,段前0.5行U三级标题字号10磅,斜体,段前0.5行Recently,variousmodelsofthehumanupper-limbanatomyhavebeenderived.Thebiomechanicalmodelsofthearmthatstandforpreciseanatomicalmodelsincludingmuscles,tendonsandbonesaretoocomplextobeutilizedinmechanicaldesignofananthropomorphicrobotarm.Fromtheviewofthemechanism,weshouldsetupamorepracticablemodelforeasyandeffectiverealization.Fig.2introducestheconfigurationofhumanupper-limbanditsequivalentmechanicalmodel,whichisa7-DOFstructure,including3degreesoffreedomforshoulder<flexion/extension,abduction/adductionandrotation>,1degreeoffreedomforelbow<flexion/extension>and3degreesoffreedomforwrist<flexion/extension,abduction/adductionandrotation>[18].ThedetailsaboutthemotioncharacteristicsoftheseskeletaljointscanbeobtainedinRefs.[18-20].Comparedtothemechanicalmodel,theshoulderandwristcanbeconsideredassphericaljointsandtheelbowasarevolutionjoint.Itisagoodapproximatemodelforthehumanarm,andthebaseforthedesignandconstructionofexoskeletonarm-ZJUESA.Fig.2.Configurationofhumanupperlimbanditsequivalentmechanicalmodel图题后遇标题时,段后回车换行2次图题后遇标题时,段后回车换行2次3.2MechanismoftheexoskeletonarmBecausethegoalofthisdeviceistofollowmotionsofthehumanarmaccuratelyforteleoperation,ZJUESAoughttomakethebestofmotionscopeofthehumanupper-limbandlimititaslittleaspossible.Aflexiblestructurewiththesameorsimilarconfigurationofhumanupper-limbisanidealchoice.Basedontheanatomyofhumanupper-limb,thejointmotionoriginatesfromextensionorflexionofthemuscleandligamentwitheachothertogeneratetorquearoundthebones.Comparedwiththeserialmechanism,themovementsoftheparallelmechanismaredrivenbytheprismatics,whichactanalogicallytothehumanmusclesandligament.Besides,usingtheparallelmechanismnotonlyrealizesthemulti-DOFjointforacompactstructureandligament.Besides,usingtheparallelmechanismnotonlyrealizesthemulti-DOFjointforacompactstructureofhumanupper-limb.Table后空一格,表序与表题间空两格表前段落,段后回车换行1次数学式前段落,段后回车换行1次单码页面页眉字号10.5磅,单倍行距,段后1.2磅The3RPSparallelmechanismisoneofthesimplestmechanisms.Fig.3explainstheprincipleofthe3RPSparallelmechanism.KIM,etal[11],introduceditintotheKISTdesign.Herewefollowthisconcept.Thetworevolutiondegreesoffreedomembodiedinthe3RPSareforflexion/extension,abduction/adductionatshoulder.ItsthirdtranslationdegreeoffreedomalongzaxiscanbeusedforthedimensionadjustmentofZJUESAfordifferentoperators.Theprismaticjointsareembodiedbypneumaticactuators,whicharedeployedtosupplyforcereflectivecapability.Alsodisplacementsensorsarelocatedalongwiththepneumaticactuatorsandthering-shapedjointstomeasuretheirlinearandangulardisplacements.Atelbow,acrank-slidemechanismcomposedofacylinderandlinksisutilizedforflexion/extension.Atwrist,sincetheabduction/adductionmovementissolimitedandcanbeindirectlyreachedbycombinationoftheotherjoints,wesimplifytheconfigurationbyignoringtheTable后空一格,表序与表题间空两格表前段落,段后回车换行1次数学式前段落,段后回车换行1次单码页面页眉字号10.5磅,单倍行距,段后1.2磅缩写点后空一格缩写点后空一格Fig.3.3RPSparallelmechanism图序与图题间空两格图序与图题间空两格Fig.4.Prototypeoftheexoskeletonarm-ZJUESA页码文字周围的图文框宽1.1cm,高页码文字周围的图文框宽1.1cm,高0.4cm,相对于"页面"水平距离18cm,相对于"段落"垂直距离0.4cm另行排的数学式必须居中,单倍行距,段后回车换行1次页码文字周围的图文框宽1.1cm,高0.4cm,相对于"页面"水平距离18cm,相对于"段落"垂直距离0.4cm另行排的数学式必须居中,单倍行距,段后回车换行1次Asnentionedabove,thebestdesignistomaketheworkspaceofZJUESAasfullycoverthescopeofthehumanupper-limbmotionaspossible.Weemploythe3RPSparallelmechanismfortheshoulder,whoseworkspacemainlyinfluencestheworkspaceofZJUESA.Theoptimaldesignof3RPSparallelmechanismfortheshoulderisthekeypointofZJUESAoptimaldesign.However,itisadesigningproblemwithmulti-factors,sayingthedisplacementoftheprismatics<factorA>,circumradiusratiooftheupperandlowerplatforms<factorB>,initiallengthoftheprismatics<factorC>,andtheircouplingparameters<factorA*B,A*CandB*C><Table1>andmulti另行排的数学式必须居中,单倍行距,段后回车换行1次页码文字周围的图文框宽1.1cm,高0.4cm,相对于"页面"水平距离18cm,相对于"段落"垂直距离0.4cm另行排的数学式必须居中,单倍行距,段后回车换行1次,<1>表题字号9磅,字体加粗,段后0.3行whereL0istheinitiallengthoftheprismatics,Risthecircumradiusofthelowerbasein3RPSmechanism,risthecircumradiusoftheupperbasein3RPSmechanism,istheexpectedreachableanglearoundaxis,andisther表题字号9磅,字体加粗,段后0.3行Table1.FactorsandtheirlevelsmmLevelrankABCA*BA*CB*C1600.5150－－－2800.438160－－－31000.389170－－－4－－180－－－表中字号表中字号8磅,行距固定值11Theorthogonalexperimentdesignisoutlinedbecauseoftheeasewithwhichlevelscanbeallocatedanditsefficiency.TheconceptoforthogonalexperimentdesignisdiscussedinRef.[21]toobtainparametersoptimization,findingthesettingforeachofanumberofinputparametersthatoptimizestheoutput<s>ofthedesign.Orthogonalexperimentdesignallowsadecreaseinthenumberofexperimentsperformedwithonlyslightlylessaccuracythanfullfactortesting.Theorthogonalexperimentdesignconceptcanbeusedforanycomplicatedsystembeinginvestigated,regardlessofthenatureofthesystem.Duringtheoptimization,allvariables,evencontinuousones,arethoughtofdiscrete"levels".Inanorthogonalexperimentdesign,thelevelsofeachfactorsareallocatedbyusinganorthogonalarray[22].Bydiscretizingvariablesinthisway,adesignofexperimentsisadvantageousinthatitcanreducethenumberofcombinationsandisresistanttonoiseandconclusionsvalidovertheentireregionspannedbythecontrolfactorsandtheirsetting.Table2describesanorthogonalexperimentdesignarrayfor6keyfactors[23].InthisarraythefirstcolumnimpliesthenumberoftheexperimentsandfactorsA,B,C,A*B,A*BandB*Carearbitrarilyassignedtocolumnsrespectively.FromTable2,36trialsofexperimentsareneeded,withthelevelofeachfactorforeachtrial-runindicatedinthearray.Theelementsrepresentthelevelsofeachfactors.Theverticalcolumnsrepresenttheexperimentalfactorstobestudiedusingthatarray.Eachofthecolumnscontainsseveralassignmentsateachlevelforthecorrespondingfactors.Thelevelsofthelatterthreefactorsaredependentonthoseoftheformerthreefactors.TheelementsofthecolumnIV,namelyfactorA*B,aredeterminedbytheelementsinthecolumnsI,II,andelementsofcolumnV,factorA*C,hastherelationshipwiththeelementsofcolumnsI,III,andthecolumnVI,factorB*C,liesonthecolumnsII,III.双数页码周围的图文框,相对于"页面"水平距离双数页码周围的图文框,相对于"页面"水平距离1.8cmTable2.OrthogonalexperimentdesignarrayL36for6keyfactorsExperimentNo.ABCA*BA*CB*CResultQ1111111Y12112122Y23113133Y34114144Y45121215Y56122226Y633331999Y333433291010Y343533391111Y353633491212Y36TherelationbetweencolumnIVandcolumnsI,IIisthat:iflevelofAisnandlevelofBism,thelevelofA*Bis3<n–1>+m,wheren=1,2,3andm=1,2,3.Allthecasescanbeexpressedasfollows:<1,1>1<1,2>2<1,3>3;<2,1>4<2,2>5<2,3>6;<3,1>7<3,2>8<3,3>9.ThefirstelementinthebracketrepresentsthecorrespondingleveloffactorAinTable1andthelattermeansthecorrespondinglevelofthefactorB.FactorA*Bhastotally9levels,asfactorAandfactorBhave3levels,respectively.Likewise,therelationbetweencolumnVandcolumnsI,IIIis<1,1>1<1,2>2<1,3>3<1,4>4;<2,1>5<2,2>6<2,3>7<2,4>8;<3,1>9<3,2>10<3,3>11<3,4>12.AlsotherelationbetweencolumnVIandcolumnsII,IIIis<1,1>1<1,2>2<1,3>3<1,4>4;<2,1>5<2,2>6<2,3>7<2,4>8;<3,1>9<3,2>10<3,3>11<3,4>12.Theoptimaldesigniscarriedoutaccordingtothefirstthreecolumns:<2>, <3>wherei=A,B,C,A*B,A*C,B*C;jisthenumberofirank.缩写点与后续文字间空一格ByEqs.<2>,<3>andthekinematicscalculationofthe3RPSparallelmechanism[24–35],therelationshipbetweenthetargetQandeachfactorcanbeobtained,asshowninFig.5.缩写点与后续文字间空一格量名称与量符号间空一格Fig.5.RelationbetweenlevelsoffactorsandQ量名称与量符号间空一格AccordingtotheplotsinFig.5,wecangetthesuperiorityandthedegreeoftheinfluence<sensitivity>ofeachdesignfactor.ThefactorwithbiggerextremedifferenceKi,asexpressedinEq.<3>hasmoreinfluenceonQ.Inthiscase,itcanbeconcludedthatthesensitivityofthefactorsA*BandA*CarehighandfactorsB*CandChaveweakinfluence,sinceKA*BandKA*CaremuchbiggerthanKB*CandKC.AndthesetA3B1,A2C1,A2,B1,C1,B1C1arethebestcombinationofeachfactorlevels.Butthereisaconflictwithformer3itemsinsuchaset.AstheirKihavelittledifferencesbetweeneachother,themiddlecourseischosen.Aftercompromising,wetakethelevel2offactorA,thelevel1offactorBandthelevel1offactorC,namelyd＝80mm,r/R＝0.5,L0＝150mmItisinterestingtoknowhowgoodtheresultsderivedfromtheabove36trialsare,whencomparedwithallotherpossiblecombinations.Becauseofitsmutualbalanceoforthogonalarrays,thisperformanceratiocanbeguaranteedbythetheoreminnon-parametricstatistics[13].Itpredictsthatthisoptimizationisbetterthan97.29%ofalternatives.Combinedwiththekinematicsanddynamicssimulationofthe3RPSparallelmechanismandZJUESAwithchosendesignparametersbyADAMS,weperformtheoptimaldesign.Table3indicatesthejointrangeandjointtorqueofeachjointonZJUESA.ItisapparentthatZJUESAcanalmostcovertheworkspaceofhumanupper-limbwellsothatitcanfollowthemotionofhumanoperationupper-limbwithlittleconstrain,asshowninFig.6.Table3.JointrangesandjointtorquesforeachjointonZJUESAJointonZJUESAJointrangeθ/<°>JointtorqueT/<N·m>Jointdensityρm/<kg·m–3>Flexion/extension<shoulder>－606036Abduction/adduction－506036Rotation<shoulder>－209018Flexion/extension<elbow>09028Rotation<wrist>－209013Flexion/extension<wrist>06028Abduction/adduction<wrist>Fig.6.Motionofexoskeletonarmfollowingtheoperator4HybridFuzzy-ControllerfortheForceFeedbackOnZjuesaInmaster-slavemanipulation,besidesthevisualfeedbackandman-machinesoftinterface,theforcefeedbackisanothergoodchoicetoenhancethecontrolperformance.Iftheslavefaithfullyreproducesthemastermotionsandthemasteraccuratelyfeelstheslaveforces,theoperatorcanexperiencethesameinteractionwiththeteleoperatedtasks,aswouldtheslave.Inthiswaytheteleoperationbecomesmoreintuitive.InourbilateralteleoperationsystemwithZJUESA,a6axisforce/torquesensorismountedontheendeffectoroftheslavemanipulatoranddetectstheforceandtorqueactingontheendeffectorduringperformingthework.Thisinformationistransferredtothemastersiteinrealtime.Withdynamiccalculation,thereferencesofthegeneratingforceonactuatorsofZJUESAareobtained.Hereafter,thefeelingcanbereproducedbymeansofthepneumaticsystem.Eq.<4>expressestherelationbetweentheforceandtorqueontheendeffectorandthetorquesgeneratingonthejoints:, <4>数学式下方的解释语及其他数学式,各行间单倍行距whereF—Forceandtorqueontheendeffector,数学式下方的解释语及其他数学式,各行间单倍行距, τ—Torqueoneachjoint,,J —JacobianmatrixofZJUESA.Bydividingtheforcearm,itiseasytogettothegeneratingforceonthejoints,suchasshoulderring,elbow,wristringandwrist,asexplainedbyEq.<5>:, <5>whereai<i=3,4,5,6>istheforcearmoftheshoulderring,elbow,elbowringandwristjoints,respectively.Asforthegeneratingforceoftheprismaticsonthe3RPSparallelmechanism,itcanbecalculatedasfollows[35]:, <6>where—Jacobianmatrixof3RPSparallelmechanism,—Torqueson3RPSparallelmechanism,,f3RPS—Forceon3RPSparallelmechanism.Therefore,withEqs.<5>,<6>,thetotalsevenforcereferencesareobtainedforthepneumaticsystemonZJUESA.Fig.7explainstheschemeofthepneumaticcylinder-valvesystemfortheforcefeedback.Therefore,withEqs.<5>,<6>,thetotalsevenforcereferencesareobtainedforthepneumaticsystemonZJUESA.Fig.7explainstheschemeofthepneumaticcylinder-valvesystemfortheforcefeedback.Therefore,withEqs.<5>,<6>,thetotalsevenforcereferencesareobtainedforthepneumaticsystemonZJUESA.Fig.7explainstheschemeofthepneumaticcylinder-valvesystemfortheforcefeedback.Therefore,withEqs.<5>,<6>,thetotalsevenforcereferencesareobtainedforthepneumaticsystemonZJUESA.Fig.7explainstheschemeofthepneumaticcylinder-valvesystemfortheforcefeedback.Therefore,withEqs.<5>,<6>,thetotalsevenforcereferencesareobtainedforthepneumaticsystemonZJUESA.Fig.7explainstheschemeofthepneumaticcylinder-valvesystemfortheforcefeedback.Therefore,withEqs.<5>,<6>,thetotalsevenforcereferencesareobtainedforthepneumaticsystemonZJUESA.当图题后面有注释时,图题前、后各0.3行当图题后面有注释时,图题前、后各0.3行Fig.7.Schemeofthepneumaticcylinder-valvesystemp1,v1,a1—Pressure,volumeandsectionareaofcylinderchamber1p2,v2,a2—Pressure,volumeandsectionareaofcylinderchamber2mp—Massofthepiston注释文字字号8磅,单倍行距,最后一行段后回车换行1次a注释文字字号8磅mL —MassofloadThehigh-speedon-offvalves,workingasthecommandcomponentsinthesystem,arecontrolledbythepulsewidthmodification<PWM>signalsfromthecontrolunits,respectively.Ratherthantheproportionalorservovalve,thisisaninexpensiveandwidelyusedmethodintheapplicationofpositionandforcecontrolinthepneumaticsystem[23–28].Tosimplifythecontrolalgorithm,thereisjustonevalveonworkatanymoment.Forinstance,whenaleftwardforceiswanted,thevalveV1worksandvalveV2isoutofwork.Underthiscase,wecancontrolthepressurep1inchamber1bymodifyingthePWMsignals.Chamber2connectstotheatmosphereatthattimeandthepressurep2insidethechamber2ofcylinderisabsolutelyambientpressure,andviceversa.Ateachportofthecylinder,thereisapressuresensortodetectthepressurevalueinsidethechamberfortheclose-loopcontrol.Andthethrottlevalvesareequippedforlimitingtheflowoutofthechambertoreducepistonvibrations.Inourpreviouswork,wegaveoutthespecificmathematicmodelsofthesystem,includingpneumaticcylinder,high-speedon-offvalveandconnectingtube[33]However,thepneumaticsystemisnotusuallyawelllinearcontrolsystem,becauseoftheaircompressibilityanditseffectontheflowline.Alsothehighlynonlinearflowbringstroublesintothecontrol.Theconventionalcontrollersareoftendevelopedviasimplemodelsoftheplantbehaviorthatsatisfythenecessaryassumptions,viathespeciallytuningofrelativelysimplelinearornonlinearcontrollers.Asaresult,forpressureorforcecontrolinsuchanonlinearsystem,especiallyinwhichthechamberpressurevibratesrapidly,theconventionalcontrolmethodcanhardlyhaveagoodperformance.Fortunately,theintroductionofthehybridcontrolmethodmentioned,givesoutasolutiontothisproblem.Butthetraditionaldesignofthehybridcontrollerisalwayscomplicatedandonlyavailabletotheproportionorservovalvesystem.Inoursystem,wefiguredoutakindofnovelhybridfuzzycontrolstrategyforthehigh-speedon-offvalves,whichismuchsimplerandcanberealizedbymicrocontrolunits<MCUs>inthecontributedarchitecture.Thisstrategyiscomposedoftwomainparts:afuzzycontrollerandabang-bangcontroller.Thefuzzycontrollerprovidesaformalmethodologyforrepresenting,manipulating,andimplementingaperson’sheuristicknowledgeabouthowtocontrolasystem.Itcanberegardedasanartificialdecisionmakerthatoperatesinaclosed-loopsysteminrealtimeandcanhelpthesystemtogetthecontrolinformationeitherfromahumandecisionmakerwhoperformsthecontroltaskorbyself-study,whilethebang-bangcontrollerisaddedtodrivetheresponseofthesystemmuchmorequickly.Fig.8showstheconceptoftheproposedhybridfuzzycontroller.Theconceptofmultimodeswitchingisappliedtoactivateeitherthebang-bangcontrollerorthefuzzycontrollermode.Fig.8.ConceptofthehybridfuzzycontrollerBang-bangcontrolisappliedwhentheactualoutputisfarawayfromreferencevalue.Inthismode,fasttrackingoftheoutputisimplemented.Thefuzzycontrollerisactivatedwhentheoutputisnearthesetpoint,whichneedsaccuracycontrol.Inthefuzzy-controlmode,weusepressureerroranditschangeastheinputvariablesonwhichtomakedecisions.Ontheotherhand,thewidthofthehighvoltageinonePWMperiodisdenotedastheoutputofthecontroller.Asmentionedabove,thePConmastersiteworksasthesupervisorforreal-timedisplaying,kinematicscalculationandexchangesthecontroldatawiththeslavecomputerandsoon.ForthesakeofreducingtheburdenofthemasterPC,thedistributedcontrolsystemisintroduced.EachcontrolunitcontainsaMega8MCUofATMELInc.,workingasahybridfuzzy-controllerforeachcylinderrespectively,andformsapressureclosed-loopcontrol.Thecontrollersamplesthepressureinchamberwith20kHzsamplingratebythein-builtanalog-digitalconverters.ThesecontrollerskeepincontactorgetthedifferentialpressuresignalsfromthemasterPCthroughRS232,asdepictedinFig.9.Inthismode,fasttrackingoftheoutputisimplemented.Fig.9.Distributedcontrolsystemofthemasterarm5ForceFeedbackExperimentsFig.10givesoutthesetupoftheforcefeedbackexperiments.Thesystemincludesthesoftinterface,dataacquisition,Mega8MCUexperimentboard,on-offvalves,sensorsofdisplacementandpressure,andtheoscilloscope.WechosethecylinderDSNU-10-40-PproducedbyFESTOInc.ThesoftsignalgeneratoranddataacquisitionarebothdesignedintheLabVIEW,withwhichusersmaytakeadvantageofitspowerfulgraphicalprogrammingcapability.Comparedwithotherconventionalprogrammingenvironments,themostobviousdifferenceisthatLabVIEWisagraphicalcompilerthatusesiconsinsteadoflinesoftext.Additionally,LabVIEWhasalargesetofbuilt-inmathematicalfunctionsandgraphicaldatavisualizationanddatainputobjectstypicallyfoundindataacquisitionandanalysisapplications.Fig.10.Set-upofforcefeedbackexperimentTheplotsinFig.11giveoutexperimentalresultsofthechamberpressureoutputswithstepinputsignalsononejoint.Whileatfrequencieshigherthan80Hz,forceissensedthroughtheoperator’sjoint,muscleandtendonreceptors,andtheoperatorisunabletorespondto,andlowamplitudedisturbancesatthesefrequencies.Weremovereflectedforcesignalsabove80HzbandbyfastFouriertransfer<FFT>andgetthesmoothedcurveintheplots.Oneisobtainedbyusinghybridcontrolstrategyandanotherisobtainedbyusingtraditionalfuzzycontrollerwithoutbang-bangcontroller.Althoughthesetwocurvesbothtrackthereferencewellwithverygoodamplitudematch<lessthan5%error>andafewmillisecondsmisalignmentinthetimeprofile,bycomparingthesetwocurves,itcanbefoundthattheadjusttimeofthecurvewithhybridcontrolstrategyislessthan0.03s,whichismuchlessthan0.05sofotherwithtraditionalfuzzycontroller.Itproveseffectofthehybridcontrolstrategy.Fig.11.ExperimentalresultswithastepsignalFig.12showstheresultsoftrackingasinusoidalcommander.Thisexperiment…upto5Hzfrequencysinusoidalcommandwell.Fig.12.ExperimentresultsforsinusoidalpressurecommandsAfterthen,anothertwoexperimentsarecarriedouttorealizethebilateralteleoperationwithsimplemotion,inwhichtheslavemanipulatoriscontrolledfortheshoulderabduction/adduction<themovementofaboneaway/towardthemidlineinthefrontalplane>andextension/flexionofelbow<themovementinthesagittalplane>bytheteleoperationwithZJUESA.Inthefirstexperiment,theoperatorperformstheshoulderabduction/adductionmovementwithZJUESA,whentheslaverobotfollowsandholdsuptheload.WiththeforcefeedbackonZJUESA,theoperatorh