简介:FortheARMAXsystemwithunknowncoefficientstheoptimaladaptivecontrolisdesignedsothatthefollowingrequirementsaremetsimultaneously:1)thetransferfunctionfromareferencesignaltothesystemoutputintheclosedloopequalsaprescribedrationalfunction;2)undertheconstraintmentionedin1)aquadraticlossfunctionisminimized;3)theparameterestimateisstronglyconsistent.
简介:Supplychainisacomplex,hierarchical,integrated,openanddynamicnetwork.Everynodeinthenetworkisanindependentbusinessunitthatunitesotherorganizationstodevelopitsvalue,thecompetitionandcooperationbetweentheseunitsarebasicimpetusofthedevelopmentandevolutionofthesupplychainsystem.Thecharacteristicsofsupplychainasacomplexadaptivesystemanditsmodelingarediscussedinthispaper,anduseanexampledemonstratingthefeasibilityofCASmodelinginsupplychainmanagementstudy.
简介:Inthispaper,atrustregionmethodforequalityconstrainedoptlmizationbasedonnondiferentiableexactpenaltyisproposed.Inthisalgorithin,thetrailstepischaracterizedbycomputationofitsnormalcomponentbeingseparatedfromcomputationofitstangentialcomponent,i.e.,onlythetangentialcomponentofthetrailstepisconstrainedbytrustradiuswhilethenormalcomponentandtrailstepitselfhavenoconstraints.Theothermaincharacteristicofthealgorithmisthedecisionoftrustregionradius.Here,thedecisionoftrustregionradiususestheinformationofthegradientofobjectivefunctionandreducedHessian.However,Maratoseffectwilloccurwhenweusethenondifferentiableexactpenaltyfunctionasthemeritfunction.Inordertoobtainthesuperlinearconvergenceofthealgorithm,weusethetwiceordercorrectiontechnique.Becauseofthespecialityoftheadaptivetrustregionmethod,weusetwiceordercorrectionwhenp=0(thedefinitionisasinSection2)andthisisdifferentfromthetraditionaltrustregionmethodsforequalityconstrainedopthnization.Sothecomputationofthealgorithminthispaperisreduced.Whatismore,wecanprovethatthealgorithmisgloballyandsuperlinearlyconvergent.
简介:Optimisingbothqualitativeandquantitativefactorsisakeychallengeinsolvingconstructionfinancedecisions.Thesemi-structurednatureofconstructionfinanceoptimisationproblemsprecludesconventionaloptimisationtechniques.Withadesiretoimprovetheperformanceofthecanonicalgeneticalgorithm(CCA)whichischaracterisedbystaticcrossoverandmutationprobability,andtoprovidecontractorswithaprofit-risktrade-offcurveandcashflowprediction,anadaptivegeneticalgorithm(AGA)modelisdeveloped.TenprojectsbeingundertakenbyamajorconstructionfirminHongKongwereusedascasestudiestoevaluatetheperformanceofthegeneticalgorithm(CA).TheresultsofcasestudyrevealthattheACAoutperformedtheCGAbothintermsofitsqualityofsolutionsandthecomputationaltimerequiredforacertainlevelofaccuracy.TheresultsalsoindicatethatthereisapotentialforusingtheGAformodellingfinancialdecisionsshouldbothquantitativeandqualitativefactorsbeoptimisedsimultaneously.
简介:Weconsiderinthispapertheproblemofrecursiveidentificationforstochasticsystemswhenthenoisemodeldoesnotsatisfythepositiverealconditionassociatedwithconvergenceofstandardalgorithms.Toavoidthepositiverealcondition,adaptivespectralfactorizationtechniquesareexploitedonthebasisofaclassofnon-standardtime-varyingrecursiveRiccatiequations.TheasymptoticpropertiesoftheRiccatiequationsarestudiedasacrucialsteptotheconvergenceresultsofthepaper.
简介:Itiswellknownthatthetransientbehaviorsofthetraditionaladaptivecontrolmaybeverypooringeneral,andthattheadaptivecontroldesignedbasedinswitchingbetweenmultiplemodelsisanintuitivelyappealingandpracticallyfeasibleapproachtoimprovethetransientperformances.Inthispaper,weshallprovethatforatypicalclassoflinearsystemsdisturbedbyrandomnoises,themultiplemodelbasedleast-equares(LS)adaptiveswitchingcontrolisstatbleandconvergentandhasthesameconvergencerateasthatestablishedforthestandardleast-squares-basedself-tunningregulators.Moreover,themixedcasecombiningadativemodelswithfixedmodelsisalsoconsidered.
简介:Inthispaperwestudyurnmodel,usingsomeavailableestimatesofsuccessesprobabilities,andaddingparticleparameter,weestablishadaptivemodels.Weobtainsomestrongconvergencetheorems,ratesofconvergence,asymptoticnormalityofcomponentsintheurn,andestimates.Withtheseasymptoticalresults,weshowthattheadaptivedesignsgiveninthispaperareasymptoticallyoptimaldesigns.
简介:Stochasticadaptivecontrolisconsideredforthediscrete-timemulti-inputandmulti-outputsystemofmulti-delaywithnoiseexpressedbyanARMAprocess.TheCARIMAmodelisaspecialcaseofthesysteminquestion.Theoptimaladaptivecontrollawisgivenanditisshownthataquadraticcostfunctionisminimizedandtheclosed-loopsystemisstable.Further,whenthesystemisofminimumphase,theconvergenceratesofparameterestimatesandofthecost-functionarealsoderived.
简介:Inmobileadhocnetworks(MANET),nodesmovefreelyandthedistributionofaccessrequestschangesdynamically.Replicaallocationinsuchadynamicenvironmentisasignificantchallenge.Inthispaoer,adynamicadaptivereplicaallocationalgorithmthatcanadapttothenodesmotionisproposedtominimizethecommunicationcostofobjectaccess.Whenchangesoccurintheaccessrequestsoftheobjectorthenetworktopology,eachreplicanodecollectsaccessrequestsfromitsneighborsandmakesdecisionslocallytoexpandreplicatoneighborsortorelinquishthereplica.Thealgorithmdynamicallyadaptsthereplicaallocationschemetoalocaloptimalone.SimulationresultsshowthatouralgorithmsefficientlyreducethecommunicationcostofobjectaccessinMANETenvironment.
简介:这份报纸考虑最佳的公事包deleveraging的问题,它是在金融的一个关键问题。拿进帐号跨影响的永久、暂时的价格,作者与框限制和单身地二次的限制建立一个二次的程序。在一些假设下面,当二次的限制是活跃的时,作者给最佳的答案必须被完成的最佳的做贸易优先级和表演。进一步,作者为模型建议一个适应Lagrangian算法,在的地方piecewise二次的发现根的方法被用来更多样地发现Lagrangian。算法的集中被建立。作者也介绍一些数字结果,它显示出算法的实用性并且验证最佳的做贸易优先级。
简介:Inthispaper,theauthorsanalyzethestabilityofaclassofinterconnectedsystemswithsubsystemunmodeleddynamicsanddynamicinteractionsemployingdecentralizedadaptivecontrollersdesignedbyWen,Zhou,andWang(2008)inthepresenceofactuatorfailures.Itwillbeshownthattheglobalstabilityoftheremainingclosed-loopsystemisstillensuredandtheoutputsarealsoregulatedtozerowhensomesubsystemsbreakdown.
简介:Theproblemofconstructingamodelreferenceadaptivecontrollawforanuncertain1-dimensionalparabolicsystemisconsideredinthisarticle.Thecontrollerdesignedhereinvolvesonlytheplantstatebutnoitsderivatives.Aprioriboundsontheplant'suncertainparametersareusedtoproposeswitchinglawswhichserveasanadaptivemechanism.Theexponentialdecaytozeroofthestateerrorwithanyprescribedrateisguaranteedbychoosingacontrollerparametercorrespondingly.Numericalstudiesarealsopresentedtoillustratetheapplicabilityofthecontrollaw.
简介:High-frequencystocktrendpredictionusingmachinelearnershasraisedsubstantialinterestinliterature.Nevertheless,thereisnogoldstandardtoselecttheinputsforthelearners.Thispaperinvestigatestheapproachofadaptiveinputselection(AIS)forthetrendpredictionofhigh-frequencystockindexpriceandcomparesitwiththecommonlyuseddeterministicinputsetting(DIS)approach.TheDISapproachisimplementedthroughcomputationoftechnicalindicatorvaluesondeterministicperiodparameters.TheAISapproachselectsthemostsuitableindicatorsandtheirparametersforthetime-varyingdatasetusingfeatureselectionmethods.Twostate-of-the-artmachinelearners,supportvectormachine(SVM)andartificialneuralnetwork(ANN),areadoptedaslearningmodels.AccuracyandF-measureofSVMandANNmodelswithboththeapproachesarecomputedbasedonthehigh-frequencydataofCSI300index.TheresultssuggestthattheAISapproachusingt-statistics,informationgainandROCmethodscanachievebetterpredictionperformancethantheDISapproach.Also,theinvestmentperformanceevaluationshowsthattheAISapproachwiththesamethreefeatureselectionmethodsprovidessignificantlyhigherreturnsthantheDISapproach.
简介:这份报纸在控制结束与匹配的一般泛音骚乱考虑参数评价和一个不稳定的一个维的波浪方程的稳定。为无限维的系统的backstepping方法在适应管理者的设计被采用。这被显示出结果靠近环的系统是asymptotically稳定的。同时,当时间去无穷,估计的参数被显示对未知参数会聚。