基于本體知識三維實體幾何特征識別推理機的研究.pdf

1、同濟大學電子與信息工程學院博士學位論文基于本體知識三維實體幾何特征識別推理機的研究姓名：秦國鋒申請學位級別：博士專業(yè)：計算機應用技術指導教師：李啟炎20040601AbstractIntegrationformathematic，geometricandsymbolicmodelsisakeyprobleminintelligentcomputeraideddesign(ICAD)Artificialintelligentneuraln

2、etworkandpatternrecognitionarebeingutilizedatCAD，thesecreatemanynewfieldsfordevelopmentofICADandeffortsomebroadtheorybasisinitsnumbercomputationandsymbolinferenceThisdissertationutilizesthenetworktheorybasedonsyllogismcl

3、oseloopmodelforprediction，dicovercheckandfeedbackconstructsa3Dsolidgeometricfeaturerecognitioninferencemachineonontologyknowledge，solvingtheintegrationproblemformathematic，geometricandsymbolicmodelinICADThismethodhasimpo

4、rtantsignificanceontheoryresearchandgreatvalueonpractice。Aframeworkfortheinferencemachinewastakenforward，whichiSmadeupbyfourkeymechanismsarepre—processorpartternreconitionprocessorpostprocessorandfeedbackprocessorUtilizi

5、ngtheontologyconception，thenumeralandsymbolicknowledgeindictionsincomputerworldareformed，whicharegainedfromexperts’knowledgeinrealityworld，thenaresummarized，refinedandconcludedinlogicworldThepre—processorrecordstheproced

6、uresandcoursesof3Dsolidmodelingdesign，appliesthemetalknowledgefromthedesignexperts’ontologyknowledgeandrulesdatabase，classifiesandpredicatesthebasicfeatureof3DsolidTwoproblemforpreprocessorWasresearched；oneisthedesignont

7、ologyknowledgeindiction，anotheristheirutilizationAintegratedtheorymodelfordesignontologyknowledgeindicationbasedonsemanticnetworkandfranletheorywasputforwardtosolvetheitegrationproblemforthreemodelsAknowledgeinferencestr

8、ategyondesignontologyutilizationwassupported，integratingthesymbolicsystemandartificialneuralnetwork，constructhgaintegratedneuralnetworkonjoineddoctrineAintegrateddeductiveandinferencetheoryonrulesWasresearchedinthepartte

9、rnrecognitionprocessorconcludingconstructingknowledgedatabase，refiningtheinferencerulesandformingtheknowledgeinferencestrategyTheprocessorrecognizesallbasicfearuresof3Dsolid，gainingtheirgeometricdimensionsandconstrainrel

10、ationships，determainingtheirmachiningrulesThepostprocessorchecksthedesignThemachiningknowledgeindicationsarecomprisedbychoicesofmachinetools，fixturesandtheirmethods，toolsandtheirapproachingmethods，machiningmethods。whicha

11、reindicatedbyuntilizingtheontologyknowledgeclustersandmachiningfunctionclustersArevelatorysearchstrategictheorymodelandalgorithminstatusspaceplansthemachiningproceduresof3Dsolidautomativelyandaffortsomefeedbackmechanismt

12、opreprocessandrevisedesignThisdissertationhassolvedtheintegrationproblemforthreeheterogeneousmodels，concludingtheirCADandCAMontologyknowledgeindicationandutilization，andtookforwardainnovativetheoryandmethodforparalleland

13、cooperationinnumeraldesignandmanufacture，whichcanbeusedinengineeringdesingandmanufacture，computergraphicstimulationandSOonKeywords：Ontologyknowledge；Knowledgeindicationandutilization；ICAD；Neurainetwork；Partternmatching；I