資源描述:
《模糊屬性Petri網(wǎng)建模方法及學(xué)習(xí)模型研究-論文.pdf》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫(kù)。
1、第40卷第6期計(jì)算機(jī)工程2014年6月V.o1.40NO.6ComputerEngineeringJune20l4·人工智能及識(shí)別技術(shù)·文章編號(hào):1000—3428(2014)06—0190—05文獻(xiàn)標(biāo)識(shí)碼:A中圖分類號(hào):TP18模糊屬性Petri網(wǎng)建模方法及學(xué)習(xí)模型研究周如旗,馮嘉禮,張謙(1.廣東第二師范學(xué)院計(jì)算機(jī)科學(xué)系,廣州510303;2.上海海事大學(xué)信息工程學(xué)院,上海200135;3.華南理工大學(xué)自動(dòng)化科學(xué)與工程學(xué)院自主系統(tǒng)與網(wǎng)絡(luò)控制教育部重點(diǎn)實(shí)驗(yàn)室,廣州510640)摘要:定性映射易于表達(dá)模糊不確定性知識(shí),但其在表達(dá)人類認(rèn)知思
2、維活動(dòng)動(dòng)態(tài)特征上存在不足;模糊Petri網(wǎng)比較符合人類思維方式,但相關(guān)參數(shù)不易獲得且其自學(xué)習(xí)能力存在較大局限性。為此,提出一種模糊屬性Petri網(wǎng)(FAPN)形式定義及建模方法。在FAPN結(jié)構(gòu)中構(gòu)建定性基準(zhǔn)參數(shù)學(xué)習(xí)方法,通過(guò)定性映射定義4類變遷發(fā)生的模糊定性判斷規(guī)則和相應(yīng)變遷發(fā)生后的結(jié)果運(yùn)算公式,給出FAPN模型的推理算法和學(xué)習(xí)機(jī)制,并模擬系統(tǒng)的動(dòng)態(tài)運(yùn)行過(guò)程。分析結(jié)果表明,該方法能有效提高FAPN的學(xué)習(xí)能力,可適用于以定性判斷為特點(diǎn)的診斷系統(tǒng)。關(guān)鍵詞:模糊屬性Petri網(wǎng);定性映射;定性基準(zhǔn)變換;定性判斷規(guī)則;知識(shí)推理;機(jī)器學(xué)習(xí)Resea
3、rch0nModelingMethodandLearningModel0fFuzzyAttributePetriNet.ZHOURu.qi,F(xiàn)ENGJia.ii,ZHANGQian,(1.DepartmentofComputerScience,GuangdongUniversityofEducation,Guangzhou510303,China;.2.CollegeofInformationEngineering,ShanghaiMaritimeUniversity,Shanghai200135,China;3.KeyLaborator
4、yofAutonomousSystemsandNetworkedControl,MinistryofEducation,SchoolofAutomationScienceandEngineering,SouthChinaUniversityofTechnology,Guangzhou510640,China)[Abstract]Thequalitativemappingcanbeeasytoexpressfuzzyuncertainknowledge,butitisnotagoodrepresentationmethodofdynamic
5、characteristicsofcognitivethinkingaction.FuzzyPetrinetismoreconsistentwithhuman’Sthinkingmode,butitsparametersarenoteasytobeobtainedandithaslimitationsinself-learningability.Forthesereasons,theformalconceptandmodelingmethodofFuzzyAttributePetriNet(FAPN)aredefined.Thelearn
6、ingmethodabouttheparametersisconstructedintheFAPNstructure.Fourtypesoffuzzyqualitativejudgmentrulesandtheoperationformulasofthetransitionnodearedefinedbasedonthequalitativemapping.ThereasoningalgorithmandthelearningmethodofFAPNareproposed,whichcansimulatethedynamicprocess
7、ofthenetworksystem.Analysisresultsshowthat,theproposedmethodcanmakeFAPNhavebetterlearningability,anditisalsousefulinthediagnosissystemcharacterizedwiththequalitativejudgement.[Keywords]FuzzyAttributePetriNet(FAPN);qualitaivemapping;qualitativecriteriontransformation;quali
8、tativejudgmentrules;knowledgereasoning;machinelearningDOI:10.3969~.issn.1000—3428.2014.06.041伸縮、