简介:ThetypicalBDI(beliefdesireintention)modelofagentisnotefficientlycomputableandthestrictlogicexpressionisnoteasilyapplicabletotheAUV(autonomousunderwatervehicle)domainwithuncertainties.Inthispaper,anAUVfuzzyneuralBDImodelisproposed.Themodelisafuzzyneuralnetworkcomposedoffivelayers:input(beliefsanddesires),fuzzification,commitment,fuzzyintention,anddefuzzificationlayer.Inthemodel,thefuzzycommitmentrulesandneuralnetworkarecombinedtoformintentionsfrombeliefsanddesires.ThemodelisdemonstratedbysolvingPEG(pursuit-evasiongame),andthesimulationresultissatisfactory.
简介:Apressurizerisoneofimportantequipmentinapressurizedwaterreactorplant.Itisusedtomaintainthepressureofprimarycoolantwithinallowedrangebecausethesharpchangeofcoolantpressureaffectsthesecurityofreactor,therefor,thestudyofpressurizer'spressurecontrolmethodsisveryimportant.Inthispaper,anadaptivefuzzycontrollerispresentedforpressurecontrolofapressurizerinanuclearpowerplant.Thecontrollercanon-linetunefuzzycontrolrulesandparametersbyself-learningintheactualcontrolprocess,whichpossessesthewayofthinkinglikehumantomakeadecision.Thesimulationresultsforapressurizedwaterreactorplantshowthattheadaptivefuzzycontrollerhasoptimumandintelligentcharacteristics,whichprovethecontrolleriseffective.
简介:Allkindsofreasonsareanalysedintheoryandafaultrepositorycombinedwithlocalexpertexperiencesisestablishedaccordingtothestructureandtheoperationcharacteristicofsteamgeneratorinthispaper.Atthesametime,Kohonenalgo-rithmisusedforfaultdiagnosessystembasedonfuzzyneuralnetworks.Fuzzyarithmeticisinductedintoneuralnetworkstosolveuncertaindiagnosisinducedbyuncertainknowledge.Accordingtoitsself-associationinthecourseofdefaultdiagnosis.thesystemisprovidedwithnon-supervise,self-organizing,self-learning,andhasstrongclusterabilityandfastclustervelocity.
简介:Itisverydifficulttoestimateexactvaluesoftimeandcostofanactivityinprojectschedulingprocessbecausemanyuncertainfactors,suchasweather,productivitylevel,humanfactorsetc.,dynamicallyaffectthemduringprojectimplementationprocess.AGAs-basedfullyfuzzyoptimaltime-costtrade-offmodelispresentedbasedonfuzzysetsandgeneticalgorithms(GAs).Intihsmodelallparametersandvariablesarecharacteristicsbyfuzzynumbers.AndthenGAsisadoptedtosearchfortheoptimalsolutiontothismodel.Themethodsolvesthetime-costtrade-offproblemsunderanuncertainenvironmentandisprovedpracticablethroughagivingexampleinshipbuildingscheduling.
简介:船冰碰撞是船舶碰撞研究领域的热点之一,对冰材料的模拟是船冰碰撞的研究重点。提出一种利用理想弹塑性模型模拟的冰材料本构模型,利用半隐式图形算法计算单元塑性阶段的应力,利用Tsai-Wu屈服准则和经验失效公式用来描述冰的力学行为。利用二次开发功能,将冰材料模型嵌入LS_DYNA程序,并验证该模型的准确性和适用性。研究中针对不同局部形状的冰块与船侧碰撞场景,通过比较分析碰撞力、能量耗散等,探讨冰块的局部形状对碰撞场景的影响。研究结果表明:冰材料模型在大接触面的条件下压力与已有标准吻合较好;在不同的冰块局部形状条件下,船冰碰撞的相互作用过程不同;较钝形状的冰块表现近乎刚体,较尖锐形状的冰块较易破碎。