電子鼻動(dòng)態(tài)特征信息處理方法研究.pdf

1、西北工業(yè)大學(xué)碩士學(xué)位論文電子鼻動(dòng)態(tài)特征信息處理方法研究姓名：王米娜申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：檢測(cè)技術(shù)與自動(dòng)化裝置指導(dǎo)教師：楊建華20050301ABSTRACTABSTRACTWiththedevelopmentofinterrelatedtechnologysuchassensorstechnologydataprocessingtechnologycomputertechnologyartificialintelligencetech

2、nologysoftwareandhardwaretechnologyofparallelcomputersandthedevelopmentofindustrythedynamicdatafusiontechniqueofelectronicnosewillbetheimportantmeansforthedevelopmentofelectronicnoseinthefutureInthisthesistheprincipleand

3、essencemeansofdynamicdatafusiontechniqueareintroduced.Somecommonusedmethodsofdynamicdatapretreatmentandpatternrecognitionalgorithmaresummarized.OnthebasisofthistheMultiBPsubNetworkandthemodelofLeastSquarealgorithmfordyna

4、micdatafusioninelectronicnosearepresented.InadditionRBFNeuralNetworkisappliedfordynamicdatafusionandthemethodofmultigroupforecastinelectronicnosesystemisproposed.Inordertocomparethediscriminatingeffectofdifferentpatternr

5、ecognitionalgorithmindynamicdatafusiontechniqueasetofdetectionsystemwhichiscombinedgassensorarray.withpatternrecognitiontechnologyisdesignedandconstructed.Bythissystemexperimentsarecarriedoutwithdrinksamplesinexotericcir

6、cumstance.Thedynamicdataareobtainedintheconditionofchangelessandchangedtemperaturerespectively.QualitativerecognitionisprocessedfromthedynamicdataandresultsshowthatthecapabilityofBPnetworkisweakrelativelybecauseoflonglea

7、rningtimeanddifficultyofconvergence.ItisusuallydifficulttoidentifythecomplicatedodorsuccessfullyusinggeneralBPnetworkinelectronicnose.Ratioofrecognitionis100%whenwecombineMultiBPsubNetworkwithdynamicdatainchangelesstempe

8、rature.WhenRBFNeuralNetworkisappliedintwokindsofdynamicdatafusionratioofrecognitioncanbeenhancedfrom89.7%and70.4%to100%and95.3%respectively.Themodelofleastsquarealgorithmissimpleandtheefficiencyisrelativehigh.Theprecisen