简介:AMarkovianriskprocessisconsideredinthispaper,whichisthegener-alizationoftheclassicalriskmodel.ItisproperthatariskprocesswithlargeclaimsismodelledastheMarkovianriskmodel.Insuchamodel,theoccurrenceofclaimsisdescribedbyapointprocess{N(t)}_(t≥0)withN(t)beingthenumberofjumpsduringtheinterval(0,t]foraMarkovjumpprocess.TheruinprobabilityΨ(u)ofacompanyfacingsuchariskmodelismainlystudied.AnintegralequationsatisfiedbytheruinprobabilityfunctionΨ(u)isobtainedandtheboundsfortheconvergencerateoftheruinprobabilityΨ(u)aregivenbyusingageneralizedrenewaltechniquedevelopedinthepaper.
简介:MarkovianarrivalprocesseswereintroducedbyNeutsin1979(Neuts1979)andhavebeenusedextensivelyinthestochasticmodelingofqueueing,inventory,reliability,risk,andtelecommunicationssystems.Inthispaper,weintroduceaconstructiveapproachtodefinecontinuoustimeMarkovianarrivalprocesses.TheconstructionisbasedonPoissonprocesses,andissimpleandintuitive.SuchaconstructionmakesiteasytointerprettheparametersofMarkovianarrivalprocesses.Theconstructionalsomakesitpossibletoestablishrigorouslybasicequations,suchasKolmogorovdifferentialequations,forMarkovianarrivalprocesses,usingonlyelementarypropertiesofexponentialdistributionsandPoissonprocesses.Inaddition,theapproachcanbeusedtoconstructcontinuoustimeMarkovchainswithafinitenumberofstates
简介:
简介:这篇论文学习产品连续地被生产的一个生产系统;其说明限制为屏蔽检查被指定。在这篇论文,我们考虑双优秀特征;与在一个更低的说明下面掉落的每个优秀特征联系的不同费用限制一个上面的说明限制的以上。由于这些不同费用,期望的全部的利润将极大地取决于过程参数,一个特别过程平均数。这篇论文为在一个单个阶段的系统与双优秀特征的考虑决定最佳进程工具开发一个基于Markovian的模型。建议模型然后通过一个数字例子被说明;敏感分析被执行验证模型。结果证明最佳为两个优秀特征处理平均数在系统的性能上有重要效果。自从处理多质量特征是极其有限的文学调查表演,建议模型,结合了Markovian途径,提供唯一的贡献给这个领域。
简介:Inthispaper,globalrobuststabilityofuncertainstochasticrecurrentneuralnetworkswithMarkovianjumpingparametersisconsidered.AnovelLinearmatrixinequality(LMI)basedstabilitycriterionisobtainedtoguaranteetheasymptoticstabilityofuncertainstochasticrecurrentneuralnetworkswithMarkovianjumpingparameters.TheresultsarederivedbyusingtheLyapunovfunctionaltechnique,LipchitzconditionandS-procuture.Finally,numericalexamplesaregiventodemonstratethecorrectnessofthetheoreticalresults.Ourresultsarealsocomparedwithresultsdiscussedin[31]and[34]toshowtheeffectivenessandconservativeness.
简介:WeinvestigatetheexponentialstabilityinthemeansquaresenseforthesystemswithMarkovianswitchingandimpulseeffects.BasedonthestatisticpropertyoftheMarkovprocess,astabilitycriterionisestablished.Then,bytheparameterizationsviaafamilyofauxiliarymatrices,thedynamicaloutputfeedbackcontrollercanbesolvedviaanLMIapproach,whichmakestheclosed-loopsystemexponentiallystable.Anumericalexampleisgiventodemonstratethemethod.
简介:这份报纸为联合冲动的Markovian的一个类被奉献给稳定性的调查跳网络(CIMJRDSN)上的反应散开系统。由使用图理论,一个系统的方法被提供为CIMJRDSN构造全球Lyapunov功能。把功能和随机的分析方法基于Lyapunov,与网络的拓扑学性质联系的一些新奇稳定性原则被建立。
简介:ThispaperstudiestherobuststochasticstabilizationandrobustH∞controlforlineartime-delaysystemswithbothMarkovianjumpparametersandunknownnorm-boundedparameteruncertainties.ThisproblemcanbesolvedonthebasisofstochasticLyapunovapproachandlinearmatrixinequality(LMI)technique.SufficientconditionsfortheexistenceofstochasticstabilizationandrobustH∞statefeedbackcontrollerarepresentedintermsofasetofsolutionsofcoupledLMIs.Finally,anumericalexampleisincludedtodemonstratethepracticabilityoftheproposedmethods.
简介:纸为积极Markovian涉及积极观察员设计跳有部分已知的转变率的系统。由使用一个线性co积极的类型Lyapunov-Krasovskii函数,一个足够的条件被建议保证错误的随机的稳定性积极系统和积极观察员的存在,它在线性编程被计算。最后,一个例子被给表明主要结果的有效性。
简介:Inthispaperwediscussthediscrete,timenon--homogeneousdiscountedMarkoviandecisionprogramming,wherethestatespaceandallactionsetsarecountable.Supposethattheoptimumvaluefunctionisfinite.Wegivethenecessaryandsufficientconditionsfortheexistenceofanoptimalpolicy.Supposethattheabsolutemeanofrewardsisrelativelybounded.Wealsogivethenecessaryandsufficientconditionsfortheexistenceofanoptimalpolicy.
简介:Thispaperaddressestheproblemonsensorfaultestimationandfault-tolerantcontrolforaclassofTakagi-SugenoMarkovianjumpsystems,whicharesubjectedtosensorfaultsandpartiallyunknowntransitionrates.First,theoriginalplantisextendedtoadescriptorsystem,wheretheoriginalstatesandthesensorfaultsareassembledintothenewstatevector.Then,anovelreduced-orderobserverisdesignedfortheextendedsystemtosimultaneouslyestimatetheimmeasurablestatesandsensorfaults.Second,byusingtheestimatedstatesobtainedfromthedesignedobserver,astate-feedbackfault-tolerantcontrolstrategyisdevelopedtomaketheresultingclosed-loopcontrolsystemstochasticallystable.Basedonlinearmatrixinequalitytechnique,algorithmsarepresentedtocomputetheobservergainsandcontrolgains.Theeffectivenessoftheproposedobserverandcontrollerarevalidatedbyanumericalexampleandacomparedstudy,respectively,andthesimulationresultsrevealthattheproposedmethodcansuccessfullyestimatethesensorfaultsandguaranteethestochasticstabilityoftheresultingclosed-loopsystem.