英語(yǔ)原文FUZZYENGINEERINGINNUCLEARRESEARCHAPPLICATIONSDRUANLVANDENDURPELANDPDHONDTFUELRESEARCHUNITNUCLEARRESEARCHCENTRESCKACENBOERETANG200,B2400MOL,BELGIUMABSTRACTSECURITY,MAINTENANCE,MONITORING,DIAGNOSIS,ANDENVIRONMENTAREALLRELATEDTOHUMANSANDTHEIRSOCIETY,ANDARETHEMOSTIMPORTANTANDDIFICULTPROBLEMSOFNUCLEARENGINEERINGTHESEPROBLEMSARESOCOMPLICATEDTHATTHEYCANHARDLYBESOLVEDWITHOUTAGLOBALAPPROACHTHEREFORE,FUZZYENGINEERINGMAYBEONEOFTHEMOSTPOWERFULTOOLSAVAILABLETOUSTHISPAPERREPORTSONTHEINITIALACTIVITYOFFUZZYENGINEERINGATTHEBELGIANNUCLEARRESEARCHCENTREITILLUSTRATESTWOAPPLICATIONS,NAMELY,NUCLEAREMERGENCYDECISIONAIDINGSYSTEMS,ANDINSPECTIONOFTRANSMISSIONLINESOFNUCLEARINSTALLATIONSWITHPRELIMINARYRESULTS,THENSTATESANEWRTYDPROJECTCONCERNINGAFUZZYMODELBASEDCONTROLOFANUCLEARREACTORGIVINGMOREPRACTICALQUESTIONSLODOWITHFUZZYENGINEERINGRATHERTHANPRESENTINGSOLUTIONSTHEPAPERFINALLYEMPHASIZESTHATFUZZYENGINEERINGISNEEDEDFORTHENUCLEARRESEARCHWORLDTHROUGHTHEOVERWHELMINGRESPONSETOTHEFIRSTINTERNATIONALFLINSWORKSHOPONFUZZLOGICANDINTELLIGENTTECHNOLOGIESINNUCLEARSCIENCE,HELDSEPTEMBER1416,1994INMOL,BELGIUM1INTRODUCTIONFLINSISANACRONYMFORFUZZYLOGICANDINTELLIGENTTECHNOLOGIESINNUCLEARSCIENCEITSTARTEDASANEWRESEARCHPROJECT,LAUNCHEDINLINEWITHTHEOBJECTIVEOFTHEBELGIANNUCLEARRESEARCHCENTRESCKOCENTOGIVEYOUNGTALENTEDPEOPLETHEOPPORTUNITYTOCARRYOUTFUTUREORIENTEDRESEARCHNOWFLINSHASBECOMETHENAMEOFANEWINTERNATIONALRESEARCHFORUMAIMINGTOPROMOTETHETHEORYANDAPPLICATIONSOFFUZZYLOGICANDOTHERNOVELINTELLIGENTTECHNOLOGIESINTHEDOMAINOFNUCLEARSCIENCEANDENGINEERINGNOWADAYSSCIENTISTSHAVEMANYMODELSATTHEIRDISPOSALTOTREATINCOMPLETEANDCOMPLEXINFORMATIONUNDOUBTEDLY,FUZZYSETTHEORYISONEOFTHEMOSTWIDELYAPPLIEDOFTHESENEWMODELSATTHEUNIVERSITYOFGHENT,FORALMOSTTWENTYYEARS,FUZZYSETTHEORYANDITSAPPLICATIONSHAVEBEENEXTENSIVELYRESEARCHEDINTHECENTREFORFUZZINESSANDUNCERTAINTYMODELLINGUNDERTHEGUIDANCEOFPROFEEKERREFLINSBUILDSUPONASTRONGCOOPERATIONBETWEENSCKOCENANDTHEUNIVERSITYOFGHENTINVENTEDBYPROFLAZADEHATTHEUNIVERSITYOFCALIFORNIAATBERKELEY,USA,IN1965,THEREHAVEBEENMORETHANFIFTEENTHOUSANDPAPERSPUBLISHEDONBOTHTHEORETICALANDPRACTICALISSUESOFFUZZYSETTHEORYHOWEVER,ASPROFHJZIMMERMANNELITE,GERMANYPOINTSOUT21,FORALMOSTTWENTYYEARS,FUZZYMETHODSWERECONFINEDTOUNIVERSITIESINMOSTPARTSOFTHEWORLD,WITHTHEPOSSIBLEEXCEPTIONOFSOMEFUZZYLOGICCONTROLAPPLICATIONSTHISISPARTICULARLYTRUEINTHEFIELDOFNUCLEARSCIENCEANDTECHNOLOGIESTHEBESTKNOWNWORKINTHISAREAISPARTICULARLYTOPICAL,ASITDEALSWITHTHECHERNOBYLACCIDENT,INWHICHFUZZYHUMANRELIABILITYANALYSESINMANMACHINESYSTEMSHASBEENCONSIDERED3,4SUBSEQUENTLY,PROFTTERANOLIFE,JAPANADDRESSEDANIMPORTANTAPPLICATIONOFFUZZYLOGICTOSOMEOFTODAYSMOSTCRITICALPROBLEMSSECURITY,MAINTENANCE,MONITORING,DIAGNOSIS,ANDENVIRONMENTAREALLRELATEDTOHUMANSANDTHEIRSOCIETY,ANDARETHEMOSTIMPORTANTANDDIFFICULTPROBLEMSOFNUCLEARENGINEERINGTHESEPROBLEMSARESOCOMPLICATEDTHATTHEYCANHARDLYBESOLVEDWITHOUTAGLOBALAPPROACHTHEREFORE,FUZZYENGINEERINGMAYBEONEOFTHEMOSTPOWERFULTOOLSAVAILABLETOUSLOTHECONTINUOUSINCREASEANDREMARKABLESUCCESSOFVARIOUSAPPLICATIONSOFFUZZYLOGICPROMPTEDTHEFOUNDATIONOFTHERESEARCHFORUMFLINSATTHEENDOF1992ATPRESENTFLINSSCKOCENCONSISTSOFSEVERALENGINEERS,MOSTLYFROMNUCLEARSCIENCE,ANDSCIENTISTSWHOARECURRENTLYWORKINGONVARIOUSPROJECTSFORRESEARCHDEGREES,POSTDOCTORATESTUDIES,OROTHERRESEARCHACTIVITIESASTHENAMESUGGESTS,ITISNOTONLYLIMITEDTOFUZZYLOGIC,BUTINCLUDESOTHERINTELLIGENTTECHNOLOGIES,SUCHASNEURALNETWORKS,KNOWLEDGEBASEDEXPERTSYSTEMS,ANDCOMPLEXPROBLEMSOLVINGTECHNOLOGIESFLINSAIMSATSOLVINGMANYINTRICATEPROBLEMPERTAININGTOTHENUCLEARENVIRONMENTBYUSINGMODERNTECHNOLOGIESASADDITIONALTOOLSTHENEEDFORTHECREATIONOFAPARTICULARFORUMSUCHASFLINSISEVIDENCEDBYTHEOVERWHELMINGRESPONSETOTHECALLFORPAPERSFORTHEFIRSTINTERNATIONALFLINSWORKSHOP,HELDSEPTEMBER1416,1994INMOL,BELGIUMABOUTSEVENTYPAPERSFROMAUSTRIA,BELGIUM,CHINA,FRANCE,GERMANY,JAPAN,KOREA,MALAYSIA,THENETHERLANDS,NORWAY,RUSSIA,THEUNITEDKINGDOM,ANDTHEUNITEDSTATESOFAMERICAWEREACCEPTEDTHEYCOVERAPPLICATIONSOFFUZZYLOGICANDINTELLIGENTTECHNOLOGIESINRADIATIONPROTECTION,NUCLEARSAFETYHUMANFACTORSANDRELIABILITY,SAFEGUARDS,NUCLEARPOWERPLANTCONTROL,DECISIONMAKING,ANDNUCLEARREACTORCONTROLTHISPAPERREPORTSONTHEINITIALACTIVITYOFFUZZYENGINEERINGATTHEBELGIANNUCLEARRESEARCHCENTREWITHTWOCONCRETEAPPLICATIONS,NAMELY,NUCLEAREMERGENCYDECISIONAIDINGSYSTEM,ANDINSPECTIONOFTRANSMISSIONLINESOFNUCLEARINSTALLATIONS,THENSTATESANEWRSECONDLY,TOPROVIDEEARLYESTIMATESOFTHESOURCETERM,ANDHENCEPERMITCONSEQUENCESFARTHERAFIELDTOBEASSESSEDTHESOURCETERMMODULEHASTHETASKOFSUPPLYINGTHESYSTEMWITHALLINFORMATIONCONCERNINGTHERELEASEOFRADIONUCLIDESHOWEVER,THEEXACTRECONSTRUCTIONOFTHESOURCETERMISUPTONOWANUNRESOLVEDPROBLEMSEVERALMATHEMATICALMETHODSAREUNDERDEVELOPMENTFORATMOSPHERICDISPERSIONMODELSSOMEOFTHEMALREADYGIVERESULTSWITHACCEPTABLEACCURACYINSPECIALBUTNOTTOOCOMPLEXSITUATIONSADIRECTAPPLICATIONOFTHEMETHODSUSEDFORPROBABILITIESASSESSMENTOFACCIDENTCONSEQUENCEISNOTSUITABLE,SINCETIMECONSUMINGCALCULATIONSARENOTFEASIBLEDURINGANEMERGENCY,ANDTHENATUREOFTHEUNCERTAINTIESINALARGEFRACTIONOFTHEINPUTINFORMATIONANDMODELPREDICTIONSDIFFERSFUNDAMENTALLYTHEREFORE,THEWORKLLSEEKSANOTHERAPPROACHBASEDONFUZZYSETSANDDECISIONTHEORYTOTREATUNCERTAININFORMATIONINCOMPLETEANDINEXACTFOREXAMPLE,ANAMBIGUOUSPROBLEMOFTHECLASSIFICATIONWITHLINGUISTICVALUESSUCHASVERYGOOD,GOOD,MIDDLE,BAD,ANDVERYBAD,ACCORDINGTOTHERATIOOFPREDICTIONSANDOBSERVATIONDATA,CANBETREATEDWITHMEMBERSHIPFUNCTIONSALSOASIMPLEFUZZYALGORITHMISDESIGNEDFORTHESYSTEMTOPROVIDEAREASONABLESOLUTIONFORPRACTICALUSERS,WHICHCOULDPROPOSEDIFFERENTSOLUTIONSWITHINASHORTTIMEPERIODTHEBASICIDEAOFTHESYSTEMISBUILTONTHEFOLLOWINGSIMPLEMATHEMATICALMODEL2CEXPYZQYXPUSSXD1INTHEABOVEEQUATION,CISTHECONCENTRATIONOFRADIOACTIVEMATERIAL,WHICHISDEPENDENTOFANUNKNOWNVARIABLEOFSOURCETERMQ,ANDANOTHERUNKNOWNVARIABLEOFWINDDIRECTIONDXANDYARECOORDINATESDEPENDINGONTHEWINDDIRECTIOND,ANDTHEOTHERPHYSICALPARAMETERSSUCHASU,AND,AREGIVENINTHISSTUDYCASEINAREALSITUATION,ONECANONONEHANDOBTAINBYSZMEANSOFOBSERVATION0MEASUREDCONCENTRATIONDATADENOTEDASFORCERTAINPOINTSKMC,ANDONTHEOTHERHANDONECANCALCULATEPREDICTIONPDENOTEDAS,BY1,2KNKIJTHEMODELIFTHESOURCETERMQANDTHEWINDDIRECTIONDAREGIVENFORTHESAMEPOINTSASUSEDINTHEOBSERVATIONFOREACHPOINTK,ONEDEFINESAFUZZYSETASKF2,1|KKMIJFCWHERETHEINDEXESI,JCORRESPOND10THEGRIDSOFTHEWINDDIRECTIONANDTHESOURCETERM,THEFUZZYSETCANBEINTERPRETEDASINACERTAINPOINTK,1D,1IJEDIQQDKFTHEPREDICTIONPISTHECLOSESTTOTHEOBSERVATION0FORA11POINTS,USINGTHEDECISIONMAKINGTHEORY1,19,20ONEOBTAINSTHEFUZZYDECISIONSETDAS31NKFTHEBESTSOURCETERMQOANDTHEBESTWINDDIRECTIONDOCANBEOBTAINEDBYUSINGTHEDEFIIZZIFICATIONTECHNIQUESPRACTICALLY,ONECANSETTHEFINITEGRIDSFORWINDDIRECTIONSINCETHERANGEOFITISMOREORLESSKNOWNHOWEVERTHEREISNOIDEAABOUTTHESOURCETERMTOSETALARGERANGEOFTHESOURCEYIELDSTHECOMPUTERTIMECONSUMINGPROBLEMFORTHISSIMULATIONFORTHATREASON,RATIOSOFPREDICTIONOVEROBSERVATIONISCOMPUTED,ANDTHESITUATIONFORWHICHALLTHESERATIOSFORALLMEASUREMENTPOINTSAREASCLOSETO1ASPOSSIBLEISBEINGSEARCHEDWITHRESPECTTOTHEDIFFERENTMEMBERSHIPFUNCTIONSINTHISWAY,ONEGETSANAPPROXIMATEVALUEOFTHESOURCE,THEREFOREAMOREORLESSCORRECTRANGEOFITSEVERALEXAMPLESHAVEBEENTESTEDANDTHEMETHODPROPOSEDHERESEEMSACCEPTABLEFROMTHEPOINTOFVIEWOFTHEPRACTICE4APPLICATION2INSPECTIONOFTRANSMISSIONLINESOFNUCLEARINSTALLATIONSTHESAFETYOFANUCLEARPOWERPLANTXPPANDOFRESEARCHREACTORSHASTOBEDEALTININCREASINGLYGREATERDEPTHDURINGTHECONSTRUCTIONPHASEOFTHENPP,SEVERALSAFETYRELATEDSOLUTIONSCANBESTUDIEDANDIDEALLJIMPLEMENTEDOLDERINSTALLATIONS,ATCERTAINNUCLEARFACILITIES,REQUIREEXTRASAFETYEVALUATIONONEOFTHEASPECTSTODEALWITHISTHESAFETYOFTHESEVERALTRANSMISSIONLINESINANUCLEARINSTALLATION,FORINSTANCETHESAFETYOFCONTROL,SAFETYAGAINSTFIRE,ETCTHEBELGIANNUCLEARRESEARCHCENTREHASTODEALWITHEXISTINGINSTALLATIONS,ANDESPECIALLYTHEFIRERISKANAL3SISOFELECTRICALPOWERCABLETRAYSINTHISPARTICULARCASESEVERALQUESTIONSARISE“TOWHATEXTENTCANTHOSECABINETSPROPAGATEFIRE”,“TOWHATEXTENTCANTHOSECABLESBETHESOURCEOFFIREDUETOTHEIRPREVIOUSSERVICELIFE”,AND“WHATARETHEMECHANISMSOBSERVED”CLEARLYASTRICTANSWERTOTHOSEQUESTIONSCANNOTBEOBTAINEDSINCEITNEEDSALOTOFENGINEERINGJUDGEMENT,ANDTHEPRACTICALINSPECTIONOFTHOSECABLETRAYSINTRODUCESALOTOFSUBJECTIVEINPUTSOBVIOUSLY,ASURVEYOFALLTRANSMISSIONLINESAGAINSTSEVERALHAZARDS,INTHENUCLEARINSTALLATION,ISANENORMOUSTASKASTHEFOLLOWINGSEPARATESTEPSHAVETOBETAKEN1EACHTRANSMISSIONLINENEEDSTOBECHECKEDBYANINSPECTOR,WHOHASTOREPORTTHEOBSERVATIONSTHATAREMADEABOVEALL,THEREHASALSOTOBEACCORDANCEBETWEENSEVERALINSPECTORSTHEMSELVES2THESEOBSERVATIONSRESULTINADATABASEOFALLLINESINSPECTED,INCLUDINGTHEREMARKSMADEFOREACHLINE,INWHICHTHEDATABASEFORMSTHESTARTOFTHEREALSAFETYEVALUATION3FINALLYADECISIONONTHESAFETYOFEACHLINE,ANDMOREOVER,OFTHEGLOBALINSTALLATIONSAFETYNEEDSTOBEMADETHESUBJECTIVITYASNOTED,ANDTHEDIFFERENTQUESTIONSINVOLVED,JUSTIFYTHEUSEOFFUZZYTECHNIQUESTHEMETHODOLOGYFORTHESAFETYEVALUATIONOFTRANSMISSIONLINESFORADEEPERCASESTUDYISBASEDONFUZZYSETTHEORYITSORIGINALAPPLICATIONWASINTHEFIELDOFMEDICALDIAGNOSIS,ANDLATERONITWASSUGGESTEDTOBEAPPLIEDINTHEFIELDOFSAFETYANALYSISOFELECTRICALINSTALLATIONSFUZZYRELATIONSANDCOMPOSIONALOPERATIONSOFFUZZYRELATIONSHAVEBEENUSEDINTHISFUZZYAPPROACHFUZZYRELATIONSBETWEENTHESYMPTOMSOFFAILUREOBSERVEDINATRANSMISSIONLINEANDSAFETYLEVELSARETOBEESTIMATEDTHISWILLBEACHIEVEDFROMTHEANALYSISOFINVESTIGATEDORPUBLISHEDTECHNICALINFORMATION,ANDFROMASURVEYOFTHEVIEWSOFAGROUPOFEXPERTSTHESERELATIONSHIPSAREINCORPORATEDINTHECOMPUTERPROGRAMFIRECABFIREEVALUATIONOFCABLES,DEVELOPEDTOCALCULATETHESAFETYVALUEOFEACHLINETHISAPPROACHILLUSTRATESTHATTHEFUZZYTECHNIQUEIMPLEMENTSAFASTERMETHODOLOGYWITHACCORDINGLYLOWERCOSTSINVOLVEDTHEFUZZYAPPROACHINVOLVESTHESTUDYOFTWOFUZZYRELATIONS,ANDRESULTSINAREDUCTIONINADEGREEOFSUBJECTIVITYOFTHESAFETYEVALUATIONPROCESSTHEBASICIDEAOFTHISAPPROACHISOUTLINEDINBRIEFLETXBEASETOFSYMPTOMSOFAGIVENTRANSMISSIONLINE,YASETOFFAILURETYPES,ANDZASETOFSAFETYLEVELSANDAREDENOTEDASTWOFUZZYRELATIONSBETWEENSYMPTOMSAND1,RY2,ZFAILURETYPES,ANDBETWEENFAILURETYPESANDSAFETYLEVELS,RESPECTIVELYTHEREFOREANEWFUZZYRELATIONFORACERTAINTRANSMISSIONLINEBETWEENSYMPTOMSANDSAFETYLEVELSDENOTEDCAN,RXZBEOBTAINEDAS412,RXZYRZWHEREISACOMPOSITIONALOPERATOROFFUZZYRELATIONSIFONEGIVESANINSPECTORACHECKLISTFOREACHTRANSMISSIONLINEWITHTHEPOSSIBLEOBSERVABLESYMPTOMSX,THANONEGETSEASILYASAFETYLEVELZFOREACHTRANSMISSIONLINETHISRELATIONSHIPALSOHOLDSFORTHECLASSICALAPPROACHHOWEVER,NATURALLYTHEFUZZYAPPROACHSEEMSTOHAVEANIMPORTANTADVANTAGEOFREDUCINGTHEDEGREEOFSUBJECTIVITYOFTHEHUMANOBSERVEREASILY,ONEMAKESACHECKLISTWITHSYMPTOMSX,BUTINTHECLASSICALAPPROACHTHISRESULTSDIRECTLYINALISTWITHTOOMUCHDETAIL,THEREBYLOOSINGTHEFINALGOALNAMELYASAFETYEVALUATIONINANECONOMICMANNERHOWEVER,THEFUZZYAPPROACHTAKESAGIVENSETOFSYMPTOMSXRELATEDTOAGIVENSETOFFAILURESYDUETOTHEFACTTHATTHISRELATIONHOLDSFOREVERYINSTALLATION,THISFUZZY1,RXYRELATIONHASTOBEMADEONLYONCEADDITIONSORCHANGESCANBEMODIFIEDEASILY1,RXYTYPICALELEMENTSOFTHESETOFSYMPTOMSARENOOBVIOUSISOLATIONDAMAGE,SMALLISOLATIONDAMAGES,MANYSMALLISOLATIONDAMAGES,EXCESSIVEISOLATIONDAMAGES,NORUST,SMALLRUST,EXCESSIVERUST,NORESISTANCECHANGE,SMALLRESISTANCECHANGE,ANDEXCESSIVERESISTANCECHANGETYPICALELEMENTSOFTHESETOFFAILURETYPESARENOLOSSOFISOLATION,SMALLISOLATIONLEAKS,ISOLATIONLEAKS,NOTRANSMISSIONSIGNALCHANGE,SMALLTRANSMISSIONSIGNALCHANGES,ANDHIGHTRANSMISSIONSIGNALCHANGEASACONSEQUENCE,TYPICALELEMENTSOFTHESETOFSAFETYLEVELSAREUNSAFEAGAINSTFIRE,QUESTIONABLELYSAFEAGAINSTFIRE,ANDSAFEAGAINSTFIRETHESETOFSYMPTOMSNEEDSTOBEDESCRIBEDONAVERYCOMPREHENSIVEMANNERFORALLPOSSIBLEOBSERVABLESYMPTOMS,SUCHTHATANINSPECTORCANPOINTOUTWHATISIMPORTANTFOREACHLINEAGOODPREPARATIONOFTHERELATIONHELPSINSPECTORS,ANDFORTHEFINALSAFETYRESULTCLEARLY,1,RXYTHISRELATIONISCOMMONFOREACHINSTALLATION,IFTHESEXANDYSETAREWELLCHOSENTHERELATIONISHOWEVERTYPICALFOREACHINSTALLATIONANDKNOWNSAFETYSTUDYIFONE2,ZISCONSIDERINGTHESAFETYAGAINSTFIRE,THESETOFSAFETYLEVELSZDESCRIBETHESAFETYAGAINSTFIREINTHEINSTALLATIONTHEREFORE,THISRELATIONHASTOBEESTABLISHEDFOREACHINSTALLATIONITSELFMOREOVER,THISRELATIONCANALSODESCRIBEINTERFERENCEBETWEENSEVERALLINES,THATPREVIOUSLYMADESURVEYSINSTALLATIONDEPENDENTTHEREISOFCOURSEAQUITEIMPORTANTECONOMICALASPECTINVOLVEDINTHERELATION4WITHTHISRELATION,ONEPERFORMSTHEENORMOUSTASKOFGETTINGTHERESPECTIVERELATIONSONLYONCE,ANDTHENONLYUSESONEPROGRAMTOGETALLINFORMATIONFOREACHLINETHISNOTONLYSAVESTIMEBUTALSOMONEYINACLASSICALAPPROACH,ASALREADYMENTIONED,THESESTEPSAREMOREORLESSREPEATEDFOREVERYLINE,SUCHTHATAGLOBALVIEWOFTHEGLOBALSAFETYISMISSINGHERE,AFTERPERFORMINGTHEINSPECTIONOFEACHLINE,ONEOBTAINSFOREACHLINEASAFETYLEVEL,ANDTHENDECIDESTOREPAIRORREPLACETHEMOSTAFFECTEDLINESACCORDINGTOSAFETYREQUIREMENTS5ANEWR2THESIGNTHEDIRECTIONOFCONTROLTHESAMEARGUMENTISVALUABLEFORTHECONTROLRODSTHEMSELVSTHEUSEOFONEPARAMETER,INCORPORATINGSIGNANDAMPLITUDEOFTHEACTION,MAKESASIMPLIFICATIONANDGENERALLYAMORERELIABLECONTROLSYSTEMUNTILNOWMANYOPERATORSHAVEUSEDTHEPOSITIONOFTHECONTROLRODSASTHEMEANSOFCONTROLLINGTHEREACTORBYASIMPLEINTEGRATIONOFTHECONTROLSIGNAL,THESAMEINFORMATIONCANBETRANSMITTEDTOTHESEOPERATORSNOWADAYSTHEREISALSOACERTAINRUSHTOIMPLEMENTDIGITALCONTROLLERSFORRESEARCHREACTORS,ANDEVENFORNPPSINTHISRESPECTTHEREISALSOARATIONALEFORINCORPORATINGASYSTEMMODELINTHECONTROLLAWTHEPIDCONTROLANDPERIODGENERATEDCONTROLAREWELLKNOWNINSPITEOFTHEKNOWLEDGEOFCONTROLLAWS,THEREISSTILLRESEARCHTOPERFORMABOUTTHECONTROLLAWIMPLEMENTEDBYUSEOFTHERATEOFCHANGEOFTHEREACTIVITY,THEREACTIVITYANDTHERATEOFREDISTRIBUTIONOFTHEDELAYEDNEUTRONPRECURSORSTHEFINALRESULTINGOUTPUTISOFTHEREACTORPERIODDYNAMICPERIODEQUATIONTHISKINDOFCONTROLLAWISLESSIMPLEMENTED,BUTISVERYUSEFULINDIGITALCONTROLLERSDUETOTHEEASEOFIMPLEMENTATIONOFCONTROLFUNCTIONSTHATARENORMALLYPERFORMEDBYAHUMANCONTROLPARAMETERSSUCHASAXIALOFFSETCANBEACCOMMODATEDBYTHEKNOWLEDGEBASEOFHUMANOPERATORSAMANIFOLDOFFUZZYLOGICCONTROLLERSFLCNOWADAYSEXISTTHEUSEFULNESSOFFLCINTHESEDIGITALIMPLEMENTATIONSISFORESEENFROMTHEFLCSADVANTAGESOFGREATERSIMPLICITY,GREATERROBUSTNESS,ANDLOWERCOSTCURRENTLYTHEREARESTUDIESTAKINGPLACECONCERNINGTHETUNINGOFTHOSEFLCRULESTHETRIALANDERRORMETHODSARETIMECONSUMING,ANDOFCOURSENOTAPPLICABLEINSAFETYRELATEDPROCESSESTHESELFTUNINGBASEDONTHEESTIMATIONOFOPTIMUMVALUESOFTHECENTROIDISAMUCHBETTERMETHODBECAUSEOFTHEONLINEIMPLEMENTABLEMETHODOLOGY,ANDNOMODIFICATIONSINTHEFLCRULESARENEEDEDANOTHERAPPROACHISTHEUSEOFADAPTIVENEURALNETWORKCONTROLDUETOTHENONDETERMINISTICNATUREOFFLC,ONELIKESTOMAKETHEMMOREDETERMINISTICUSINGAMATRIXEVALUATIONBYNEURALNETWORKSANOTHERAPPROACHISTHEFLCUSINGTHEERROR,CHANGEOFERRORANDTHEPREVIOUSCONTROLINPUTTHISPIFLCGIVESABETTERPERFORMANCEINSTEADYSTATETHANACONVENTIONALPIFLCHOWEVER,THENECESSARYWEIGHTINGFUNCTIONSAREDIFFICULTTODETERMINEWITHOUTUSINGINFORMATIONABOUTTHEPLANTTOBECONTROLLEDTHEQUESTIONARISES,”WHYTOUSEMODELBASEDCONTROLLAWS”,PERIODGENERATEDCONTROLCANBEUSEDFORNONLINEARSYSTEMS,ANDINTHECASEOFRATECONSTRAINEDSYSTEMS,THERESULTINGCONTROLACTIONAPPROACHESTIMEOPTIMALBEHAVIOURBUTTHENONLINEARPROBLEMHASTOBESOLVEDBYUSEOFAMODELOFTHESYSTEMDYNAMICSEXISTINGMETHODSAREHOWEVERCOMPUTATIONINTENSIVETHETIMEREQUIREDTOCALCULATETHECONTROLACTIONSTHATCORRESPONDSTOTHEOPTIMALTRAJECTORYMAYEXCEEDTHATAVAILABLEFORIMPLEMENTINGTHEASSOCIATEDCONTROLSIGNALTHEREFORE,ONESHOULDHAVETOGOOFFLINE,RESULTINGINNOFEEDBACKOFTHECONTROLLEDSYSTEMITSELFIFASYSTEMMODELISUSEDWHICHISINACCURATEORIFPERTURBATIONOCCURS,THERESULTINGRESPONSEWILLNOTBEASDESIREDTHEDEGREETOWHICHAPERIODGENERATEDCONTROLLAWAPPROACHESATIMEOPTIMALRESPONSEDEPENDSONTHETREATMENTOFTHEACCELERATIONTERMMODELBASEDCONTROLLAWSOFFERTHEPOSSIBILITYOFREPLICATINGCERTAINFUNCTIONSTHATHUMANSPERFORMDURINGTHECOURSEOFCONTROLLINGAPROCESSFINALLY,DIAGNOSISREMAINSTHERESPONSIBILITYOFTHELICENSEDOPERATORS52THECONCEPTOFSAFETYANIMPORTANTCONCEPTINTHEFUZZYMODELBASEDCONTROLOFANUCLEARREACTORISTHEINTEGRATIONOFTHECONCEPTOFSAFETYCLEARLYTHECONTROLOFANUCLEARREACTORINVOLVESAPARTOFACOMPLEXPROCESSTHESAFETYOFSUCHPROCESSESISOFCOURSEOFPRIMARYCONCERNTHEPROCESSOFREINFORCEMENTLEARNINGANDQLEARNINGCANDEALWITHTHISKINDOFCONTROLAVIABILITYDOMAINDETERMINESTHECONSTRAINTSTHATTHESTATEOFTHESYSTEMMUSTOBEYATEACHINSTANTTHEAPPLICATIONOFAVIABILITYCONSTRAINTCANRELEASETHEOBJECTIVETOFOLLOWAREFERENTIALTRAJECTORYSUCHASTIMEOPTIMALLOADFOLLOWINGFORPROCESSESWHICHARERATHERTOOCOMPLEXINTHESECASES,THECONTROLOFAPROCESSTOASAFESTATEINVOLVESASTATEWHERETHECONTROLLEDSYSTEMISNOTVERYSENSITIVEFORDISTURBANCESANDCONSEQUENTLYDOESNOTEASILYGOOUTOFCONTROLTHEREINFORCEMENTLEARNINGALGORITHMSDONOTNEEDASIGNEDERRORASINOTHERALGORITHMS,ITONLYREQUIRESASCALARREINFORCEMENTSIGNALASPERFORMANCEFEEDBACKFROMTHEENVIRONMENTFORINSTANCE,THEEXPERTSCANCOPEWITHTHEQUESTION,“WHICHCONTROLACTIONISSAFER“,ANDINTHISRESPECTFORCETHECONTROLLERTOFOLLOWTHISCONTROLACTIONTHEOBJECTIVEOFREINFORCEMENTLEARNINGISTODISCOVERACONTROLPOLICYTHATMAXIMISETHEOVERALLREINFORCEMENTITRECEIVESTHEREFORE,REINFORCEMENTLEARNINGTECHNIQUESREQUIREAMETHODFORESTIMATINGTHELONGTERMCONSEQUENCESOFACTIONSASPECIFICTECHNIQUEINVOLVESQLEARNING18THISISAMETHODFORONLINEADAPTIVECONTROLANDISWELLSUITEDTOSOLVEOPTIMALCONTROLPROBLEMSITDOESNOTUSEANEXPLICITMODELOFTHEDYNAMICALSYSTEM,IEITISADIRECTMETHODTHEUSEOFAFUZZYCONTROLMODELANDTHENOTIONSAFETYREFLECTSINASEARCHFORSAFEREGIONSWITHINTHEVIABILITYDOMAIN,ANDTODESIGNTHECONTROLLERSUCHTHATTHESYSTEMISDRIVENINTOASAFEREGIONTHEDRIVINGFORCEISDEALTWITHBYSUPERVISORYLEARNINGOFTHECONTROLLERTHEFUZZYCONTROLLERMODELISDEFINEDASAFUZZYFUNCTIONFWHICHMAPSAFUZZYREGION,MODELLEDBYAFUZZYSETOVERTHESTATESPACE,I,TOASPECTRUMOFCONTROLACTIONS,MODELLEDBYAFUZZYSETOVERTHEPOSSIBLECONTROLACTIONS,0,THEIMAGEFI,OFEACHFUZZYREGIONI,EXPRESSESAFUZZYRESTRICTION0,ONTHEPOSSIBLECONTROLACTIONSFORTHATREGIONTHESAFETYOFAFUZZYREGIONISREFLECTEDINTHESPECTRUMOFTHECORRESPONDINGIMAGETHESAFETYOFAFUZZYREGIONI3ISMEASUREDBYTAKINGTHEINTEGRALOFTHEMEMBERSHIPFUNCTIONOFTHECORRESPONDINGIMAGEJO5AJJSIUOADWHEREAISTHESETOFCONTROLACTIONSDURINGTHELEARNINGPERIODONEALSONEEDSTOMEASURETHESAFETYOFACRISPSTATESINCEACRISPSTATEGENERALLYBELONGSTOSEVERALFUZZYREGIONS,THESAFETYMEASUREISDEFINEDASTHEWEIGHTEDAVERAGEOFTHESAFETYMEASURESOFALLTHEFUZZYREGIONSTOWHICHTHECRISPSTATEBELONGSTHISKINDOFAPPROACHTOBRINGINTHECONCEPTOFSAFETYISVERYIMPORTANTINSEVERALWAYSITISNOTUNCOMMONTHATTHESAFECONTROLOFAREACTORISBEINGREPRESENTEDASATRAJECTORYINTHESTATEPLANEEARLIERMETHODSINCONVENTIONALCONTROLTHEORYMADEEXTENSIVEUSEOFTHISREPRESENTATIONINTHEMEANTIME,SUCHMETHODSAREALSOACCEPTEDASVALUABLEANDCORRECTMETHODSTOTUNECONTROLLERSTHECONCEPTOFSAFETY