简介：与超级计算机的快速的开发，规模和复杂性曾经正在增加，并且可靠性和跳回面临更大的挑战。在容错有许多重要技术，例如基于差错预言的积极失败回避技术，反应容错基于检查点，和安排技术到改进可靠性。系统差错的特征上的质、量的描述为这些技术是很批评的。这研究在超级计算机把SunwayBlueLight称为的二典型petascale上分析失败的来源(基于多核心中央处理器)并且SunwayTaihuLight(基于异构的manycore中央处理器)。它揭开一些有趣的差错特征并且在主要部件差错之中发现未知关联关系。最后，纸在资源和不同时间跨度的各种各样的谷物分析二台超级计算机的失败时间，并且为petascale超级计算机造一个一致多维的失败时间模型。

简介：Inthispaper,theoptimalmaintenancepolicyisinvestigatedforasystemwithstochasticleadtimeandtwotypesoffailures.Thesystemhastwotypesoffailures,onetypeisrepairable,whentherepairablefailureoccurs,thesystemwillberepairedbyrepairman,andthesystemafterrepairisnot“asgoodasnew".Theothertypeoffailureisunrepairable,andwhentheunrepairablefailureoccursthesystemmustbereplacedbyanewandidenticalone.Thesparesystemforreplacementisavailableonlybyorder,andtheleadtimefordeliveringthesparesystemisstochastic.Thesuccessivesurvivaltimesofthesystemformastochasticallydecreasinggeometricprocess,theconsecutiverepairtimesafterfailuresofthesystemformarenewalprocess.Byusingtherenewalprocesstheoryandgeometricprocesstheory,theexplicitexpressionofthelong-runaveragecostperunittimeunderorderingpolicy(N-1)isderived,andthecorrespondingoptimalcanbefoundanalytically.Finally,thenumericalanalysesaregiven.

简介：

简介：ThispaperisconcernedwiththeanalysisofafeedbackM[X]/G/1retrialqueuewithstartingfailuresandgeneralretrialtimes.Inabatch,eachindividualcustomerissubjecttoacontroladmissionpolicyuponarrival.Iftheserverisidle,oneofthecustomersadmittedtothesystemmaystartitsserviceandtherestjoinstheretrialgroup,whereasalltheadmittedcustomersgototheretrialgroupwhentheserverisunavailableuponarrival.Anarrivingcustomer(primaryorretrial)mustturn-ontheserver,whichtakesnegligibletime.Iftheserverisstartedsuccessfully(withacertainprobability),thecustomergetsserviceimmediately.Otherwise,therepairfortheservercommencesimmediatelyandthecustomermustleavefortheorbitandmakearetrialatalatertime.ItisassumedthatthecustomerswhofindtheserverunavailablearequeuedintheorbitinaccordancewithanFCFSdisciplineandonlythecustomerattheheadofthequeueisallowedforaccesstotheserver.TheMarkovchainunderlyingtheconsideredqueueingsystemisstudiedandthenecessaryandsufficientconditionforthesystemtobestableispresented.Explicitformulaeforthestationarydistributionandsomeperformancemeasuresofthesysteminsteady-stateareobtained.Finally,somenumericalexamplesarepresentedtoillustratetheinfluenceoftheparametersonseveralperformancecharacteristics.

简介：Theformationoflandslidedamsisofteninducedbyearthquakesinmountainousareas.Thefailureofalandslidedamtypicallyresultsincatastrophicflashfloodsordebrisflowsdownstream.Significantattentionhasbeengiventotheprocessesandmechanismsinvolvedinthefailureofindividuallandslidedams.However,theprocessesleadingtodominofailuresofmultiplelandslidedamsremainunclear.Inthisstudy,experimentaltestswerecarriedouttoinvestigatethedominofailureoflandslidedamsandtheconsequentenlargementofdownstreamdebrisflows.Differentblockageconditionswereconsidered,includingcompleteblockage,partialblockageanderodiblebed(noblockage).Themeanvelocityoftheflowfrontwasestimatedbyvideos.Totalstresstransducers(TSTs)andLaserrangefinders(LRFs)wereemployedtomeasurethetotalstressandthedepthoftheflowfront,respectively.Underacompleteblockagepattern,aportionofthedebrisflowwastrappedinfrontofeachretainedlandslidedambeforethelattercollapsedcompletely.Thiswasaccompaniedbyadramaticdecreaseinthemeanvelocityoftheflowfront.Conversely,underbothpartialblockageanderodiblebedconditions,themeanvelocityoftheflowfrontincreasedgraduallydownwardalongtheslopingchannel.Dominofailuresofthelandslidedamsweretriggeredwhenaseriesofdams(completeblockageandpartialblockage)weredistributedalongtheflume.However,notallofthesedominofailuresledtoenlargeddebrisflows.Themodesofdamfailureshavesignificantimpactsontheenlargementofdebrisflows.Therefore,furtherresearchisnecessarytounderstandthemechanismsofdominofailuresoflandslidedamsandtheireffectsontheenlargementofdebrisflows.

简介：Inthispaper,theauthorsanalyzethestabilityofaclassofinterconnectedsystemswithsubsystemunmodeleddynamicsanddynamicinteractionsemployingdecentralizedadaptivecontrollersdesignedbyWen,Zhou,andWang(2008)inthepresenceofactuatorfailures.Itwillbeshownthattheglobalstabilityoftheremainingclosed-loopsystemisstillensuredandtheoutputsarealsoregulatedtozerowhensomesubsystemsbreakdown.

简介：ConstructedwetlandwasfirstintroducedintotheUnitedKingdominthemiddleof1980s,followingavisitbyagroupofscientisttoWesternGermany.Inthepast2decades,theapplicationsofconstructedwetlandsinthiscountryhaveexpandedsubstantially,duetothedemandforgreentechnologiesandrisingcostoffossilfuelenergies.Thispaperreportedastatisticalinvestigationoftheperformancesof78horizontalflowwetlands,representativesofsuchsystemintheUnitedKingdom.Alternativedesignequations,basedonorganicmatterremovalefficiency,havebeendevelopedfromMonodkinetics,andtheaccuracyandreliabilityofcurrentandalternativedesignapproacheshavebeenexamined.

简介：在挑战性的环境，感觉数据必须在水池失败的情况下在网络内被存储，我们需要与可得到的存储空间和剩余精力从弄空的存储来源节点再分配溢出的数据项目到传感器节点。我们设计有效数据存储算法与优先级说出分布式的数据保藏的一个分布式的精力(D2P2)。这个算法考虑两数据再分配费用和数据检索费用并且把这二个问题合为一个单个问题。D2P2能有效地由在传感器节点之中使用合作通讯认识到数据再分配。以便解决再分配竞争问题，我们介绍数据优先级的概念，它能避免在来源节点和还原剂精力消费之间的竞争咨询。最后，我们由理论和模拟验证建议算法的表演。我们表明那D2P2s性能以数据保藏时间以精力消费和表演优势接近最佳的集中的算法。

简介：批评基础结构正在变得对串联失败逐渐地相互依赖、脆弱。存在研究分析了相互依赖的网络的危险到从网络拓扑学结构的静态的观点串联失败。这份报纸开发在力量网络认为负担的动态再分配探索相互依赖的力量水网络的危险的一个更现实主义的串联失败模型。在这个模型，批评忍耐阀值原来被建议显示网络的危险到串联失败。另外，是重要的联网危险的一些关键参数通过数字模拟被识别并且确定。当忍耐参数的值在批评忍耐阀值上面时，结果证明串联失败能被阻止。否则相互依赖的网络在攻击力量节点的批评部分以后崩溃。相互依赖的网络随互相依赖力量的增加变得更脆弱，它暗示保护那些互连的节点减少串联失败的后果的重要性。相互依赖的网络在高负担的攻击下面是很脆弱的，它显示出保护高负担的节点的意义。

简介：普洛斯彼罗和浮士德作为《暴风雨》和《浮士德博士》中的主角，在英国文艺复兴文学史上分别代表着好魔法师和坏魔法师。普洛斯彼罗的普魔法与浮士德追求的恶魔法大相径庭。浮士德直到最后一刻依然表现出对魔法的狂热也与普洛斯彼罗在最后一幕里平静地放弃魔法截然不同。作为一个恶魔法的实施者，浮士德博士为了追求他所谓的魔法师哲学而放弃了神学研究，进入了邪恶的世界，他实施的是“黑魔法”，最终必将会“堕入万劫不复的深渊”。然而普洛斯彼罗的魔法总的来说却是好的，与高尚的学术和精神生活是相关的。他的魔法只属于“从米兰放逐的时期”。是局限在孤岛之际的，跟米兰无关。普洛斯彼罗最终放弃了魔法，因为魔法已经完成了其应尽的义务，而对于其现在所追求的目标已不再有益。然而，从魔法的完成情况来看：普洛斯彼罗和浮士德都没有完成一个真正的魔法师所应尽的责任。前者出于自愿放弃了自己的魔力，后者则是被迫放弃了他倾其一生追求的魔力。

简介：

简介：Inthispaper,indirectadaptivestatefeedbackcontrolschemesaredevelopedtosolvetherobustfault-tolerantcontrol(FTC)designproblemofactuatorfaultandperturbationcompensationsforlineartime-invariantsystems.Amoregeneralandpracticalmodelofactuatorfaultsispresented.Whilebotheventualfaultsonactuatorsandperturbationsareunknown,theadaptiveschemesareaddressedtoestimatethelowerandupperboundsofactuator-stuckfaultsandperturbationsonline,aswellastoestimatecontroleffectivenessonactuators.Thus,onthebasisoftheinformationfromadaptiveschemes,anadaptiverobuststatefeed-backcontrollerisdesignedtocompensatetheeffectsoffaultsandperturbationsautomatically.AccordingtoLyapunovstabilitytheory,itisshownthattherobustadaptiveclosed-loopsystemscanbeensuredtobeasymptoticallystableundertheinfluenceofactuatorfaultsandboundedperturbations.Anexampleisprovidedtofurtherillustratethefaultcompensationeffectiveness.

简介：摘要：本文通过对一起站内稳控主从机之间光纤衰耗大 ,引起的通道间歇性异常报警及一起线路保护复用式光纤通道 ,因 2M线损坏引起的线路保护光纤通道中断处理案列，给出具体的解决处理方案，同时对光纤通道的组建、运行维护提出了建议。