船舶避碰的自適應模糊專家系統(tǒng)研究.pdf

1、華中科技大學博士學位論文船舶避碰的自適應模糊專家系統(tǒng)研究姓名：朱曉琳申請學位級別：博士專業(yè)：船舶與海洋結構物設計制造指導教師：許漢珍2001.4.1華中科技大學博士學位論文ABSTRACTThepurposeofthispaperiStodevelopacollision—avoidanceexpertsystemapplicabletotheshipnavigationinrealencountersandthesimulateden

2、counter—manoeuvreintrainingBasedonthemethod‘ofadaptivefuzzysystemandneuralnetworks，combinedwiththesimulationofshipmanoeuvring，anadaptivefuzzyexpertsystemfor’collisionavoidanceofships(AFESCA)iSbuiltThesystemhasfacultiesof

3、adaptivelearning，fuzzyassociativememoryandparallelfuzzyinferringAnditcancarryoutnavigationforcollisionavoidanceanddynamicsimulationinareal—timecircumstanceFirstinthestudiesofriskassessment，asubjectiveshipdomainmodelbased

4、onneuralnetworksiSpresentedwhichconsiderstheeffectsofshipmanoeuvrabiIityandvisibilityAndthen，usingthemethodoffuzzysystem，amodelofriskassessmentiSbuiltAtdif詫rentstagesofanencountermanoeuvre，asubiectivedomainofdifferentobj

5、ectsisusedinriskassessmentSecondtheknowledgerepresentation，acquirementandadaptive1earningarestudiedCollision—avoidanceknowledgeiSrepresentedinfuzzyassociativememory(FAMlrulesandiSacquiredintheprocessesofreal—timesimulate

6、dencountermanoeuvresThedifferentialcompetitivelearningalgorithmofneuralnetworksiSusedtoadaptivelygenerateFAMrulesfromknowledgesamplesInthiswaythesystemiSabletorepresentthecomplicatedexpert、knowledgeinasimplenumericalfram

7、eworkandlearntheunderlyingknowledgebehindtheexpertbehaviors：Meanwhile，aimedatparallelfuzzyinferenceinrealtime，abinaryinputoutputFAMsystemiSbuiltasaninferenceengineItiSdemonstratedtobeabletomeettheneedforreal—timenavigati

8、onBesides，thesimulationofshipmanoeuvreinstillwaterandwavesarestudiedwhichiSusedtothemanoeuvreeffectprediction，thecollision—avoidancedecisionoptimizationandt11ereal—timesimulatednavigationItenablesAFESCAto、presentoptimize