重要----模型预测控制(席裕庚)

393Vol.39,No.320133ACTAAUTOMATICASINICAMarch,2013|1;21;21;230,(Modelpredictivecontrol,MPC),,.,,.,.,,,,,.|.,2013,39(3):222¡236DOI10.3724/SP.J.1004.2013.00222ModelPredictiveControl|StatusandChallengesXIYu-Geng1;2LIDe-Wei1;2LINShu1;2AbstractSincelast30yearsthetheoryandtechnologyofmodelpredictivecontrol(MPC)havebeendevelopedrapidly.However,facingtotheincreasingrequirementsontheconstrainedoptimizationcontrolarisingfromtherapiddevelopmentofeconomyandsociety,thecurrentMPCtheoryandtechnologyarestillfacedwithgreatchallenges.Inthispaper,thedevelopmentofMPCtheoryandindustrialapplicationsisbrie°yreviewedandthelimitationsofcurrentMPCtheoryandtechnologyareanalyzed.ThenecessitytostrengthentheMPCresearcharoundscienti¯city,e®ectiveness,applicabilityandnonlinearityispointedout.Webrie°ysummarizerecentdevelopmentsandnewtrendsintheareaofMPCtheoreticalstudyandapplications,andpointoutthattostudytheMPCforlargescalesystems,fastsystems,lowcostsystemsandnonlinearsystems,willbesigni¯cantforfurtherdevelopmentofMPCtheoryandbroadeningMPCapplication¯elds.KeywordsModelpredictivecontrol(MPC),constrainedcontrol,largescalesystem,nonlinearsystemsCitationYu-GengXi,De-WeiLi,ShuLin.Modelpredictivecontrol|statusandchallenges.ActaAutomaticaSinica,2013,39(3):222¡236(Modelpredictivecontrol,MPC)70,[1¡3].,30,.,,,2012-06-252012-09-29ManuscriptreceivedJune25,2012;acceptedSeptember29,2012RecommendedbyAcademicianHUANGLin(60934007,61074060,61104078)SupportedbyNationalNaturalScienceFoundationofChina(60934007,61074060,61104078)1.2002402.()2002401.DepartmentofAutomation,ShanghaiJiaoTongUniversity,Shanghai2002402.KeyLaboratoryofSystemControlandInformationProcessingofMinistryofEducation(ShanghaiJiaoTongUniversity),Shanghai200240ActaAutomaticaSinica,vol.39,no.3,pp.222¡236,2013.,.,,,()(),.,,,[4][5][6][7][8][9][10],,.,,,.3:|223170,,.,[1¡3;11¡12],,,.,.,,[1].Qin2003[1],19995,4600,,.,\,.,.,,:1),,,,,,(DCS).Qin[1],2942,,1985550144.1999,,,.,,,.2),,,,[1],2%,.,,,.,,.,.,NMPC05NMPC08,[13¡14].,,[15].3),(Ad-hoc),,,,,,,.30,,.,,AspenTechnology22439,[16].,,,,,,.,,.,.,.,[17]:8090,90.,,,.,,,.,,,,.,.1),,,,.,(),[18],,,,,.,,,Lyapunov,,.,,,,,.2),,(Linearmatrixinequality,LMI),.,,,.,,.,LMI,,LMI,.,,,.,,.,,,,,.,,,,.:1).,,3:|225.2).,,,.3).,,PID\.4).,,.2,,,.,,.2007»2011,ElsevierIEEE,1319,Au-tomaticaControlEngineeringPracticeJournalofProcessControlIEEETransactionsonAuto-maticControl747516435.20082011IFAC,131138.\Asurveyofindustrialmodelpredictivecontroltechnology[1]2008IFACCEP,\Constrainedmodelpredic-tivecontrol:stabilityandoptimality[2]2011IFAC(HighImpactAward).,[19¡26],[27][28][29][30][31][32][33][34],.,,,,;,,,.,,,.,:1),.,.a):,[7;35¡38],,.,,.,,[39],,,.,[40¡41],,,.[42].,,,[43][44][45][46].b):/22639,,,,.,\,,.[47][18];[48],;[49];[50],Setmembership,.Bemporad(Explicit)[51¡52],,,;,,.,,,,,[53¡54].NP-hard,,,[55].,.[56](LatticePWAfunction),;[57],,;[58],;[59].,,,.,,\[1],\(Blocking)[60][61],,,,.,.,..Blocking,[62],[63¡64],.[65],,,,.,[66]/.c):,,.[3](Quadraticprogramming,QP).[55]Newton-Raphson(Semi-de¯niteprogramming,SDP),10.[67]QP.[68]\,,,,,[69].[70].,,,[71].2)90,,,.3:|227a),,.,,.,[72]Tube,;[73],,;,[74¡75],[75],.[75],[76].b),[17].,,.,.,.Mayne[2],,,,.[77][78¡79],[80].,[81¡82],,,.LPV[83].,[84],,;[85¡86],,,[87],;,[88],,;[89]H1,.3),.,,.a)90[2],[90].,,[13¡14],,.,T-S.[91]T-S,,LMI,;[92]T-S,O®-setfree,,[93¡94].,.Bem-porad1999[95],,[96¡97].[98¡99].:,,(),.b),22839,,..,.[100],,.[101],,.,Tube[102¡104].,Markov[105¡106],,.,[107¡108],,,,,,.[109],,[110],.4),,,.a),,.,.,(EconomicMPC)[111];,,(Real-timeoptimization,RTO),RTO,RTO,RTO,[112].[113]Dantzig-Wolfe.,,,,.,.[114],,[115],PTA,;[116][117].,[118],[119],[120],(),()()[121¡123],.b),,,.20089,103\,,,,[124].,[125],[126],[127],[128],[129¡130].,.,,.[131];[132],Lyapunov;[133]/.c)3:|229,,,[134];,,[135];[136];[137¡138].,[139].,[140],[141],,,(SMART)[142¡144],/[145¡146]..,.,[147],[148].,,[149¡151].,[152][153¡154][155][156][157],.5),,,.(PLC),[158]PLC,,,,PLC.,FPGA.[159],,,;,[160],;[161],,FPGA,;[162],[163].,,,,.3,,,.,,,,,,.1.1),,,.,,,:a)230391Fig.1Therelationshipbetweenfourtypicalsystemsandapplication¯elds/existingproblems(),,,.b),,,.c),,.d),,,.,.2),,,,,.:a),,,,.b),,,,,,.3),.,,PID,PID,,PID.,:a),,,PID,,.b),,.c),,,.,,,.4),.,,,.3:|231,.,:a),,,.b),,.,,.NP-hard,,,,,.,,.,,.,,,,,,,.References1QinSJ,BadgwellTA.Asurveyofindustrialmodelpredic-tivecontroltechnology.ControlEngineeringPractice,2003,11(7):733¡7642MayneDQ,RawlingsJB,RaoCV,ScokaertPOM.Con-strainedmodelpredictivecontrol:stabilityandoptimality.Automatica,2000,36(6):789¡8143MorariM,Bari¶cM.Recentdevelopmentsinthecontrolofconstrainedhybridsystems.ComputersandChemicalEn-gineering,2006,30(10¡12):1619¡16314WangWL,RiveraDE,KempfKG.Modelpredictivecon-trolstrategiesforsupplychainmanagementinsemiconduc-tormanufacturing.InternationalJournalofProductionEco-nomics,2007,107(1):56¡775DufourP,MichaudDJ,Tour¶eY,DhurjatiPS.Apartialdi®erentialequationmodelpredictivecontrolstrategy:ap-plicationtoautoclavecompositeprocessing.ComputersandChemicalEngineering,2004,28(4):545¡5566SalsburyT,MhaskarP,QinSJ.Predictivecontrolmethodstoimproveenergye±ciencyandreducede