基于結構與組份的木材品質(zhì)神經(jīng)網(wǎng)絡分析模型.pdf

1、東北大學碩士學位論文基于結構與組份的木材品質(zhì)神經(jīng)網(wǎng)絡分析模型姓名：趙亮申請學位級別：碩士專業(yè)：控制理論與控制工程指導教師：井元偉20041201東北大學碩士學位論文ABSTRACTWoodPropertyAnalysisNeuralNetworkModelBasedonStructureandComponentAbstractW油thedevelopmentofwooddirectionalcultivationtransgenicen

2、gineeringandwoodqualityreformativetechnology，theresearchofwoodqualityandformingmechanismhasbecomethefocusofexpertsandacademiciansinthefieldofwoodindustryandengineeringworldwideMoreover，itisingreatneedofeducingitsexternal

3、phv’sicalandmechanicalqualityfromwoodinternalmicrostructureandchemicalcomponentHowever，itisvirtuallydifficulttokeepsuchakindofcomplicatedwoodsystemsinhandalthoughsomanypeoplehavebeendedicatedtothat，andtraditionalresearch

4、wasconfinedtoseveralparametersandunabletoworkoutaccuratemodel_Inthisthesis，neuralnetwork(NN)theoryisintroducedintomodelingtasksBasedonNNmodelingmethod，aseriesofrelationalmodelsofwoodmicrostructure，chemicalcomponentandwoo

5、dpropertyarepresented，whichbreakthroughthetraditionalresearchapproachTheproposedmodelsherefindouttheinternalrelationshipbetweenintemalandexternalstructuresofwoodroundlyandaccuratelyRelevantsoltwareisimplementedwithVisual

6、CandMATLABwhichissupposedtobepopularizedinwoodresearchdepartmentUpondemandanalysisofthenational“973”climbingprojectofChineseAcademyofForestryakindofNNmodelnamedGeneralizedRegressionNeuralNetwork(GRNN)isselectedFirstly，th

7、erelationmodelsareestablishedfollowingtheanalysisandpreprocessofexperimentaldataandtheperformanceofNNisassessedbylinearregressionSecondly，accordingtosampledistributingcharacteristic，thedataaregroupedbygrowthringTheradial

8、variationisreflectedandthedesiredapproximationdegreeisachievedinthesub—modelsFurther，acertainmathematicalmethodisappliedtoanalyzeleader—followerinputsInthiswayaseriesofreducedmodelsareprovidedTotheaccomplishedmodelingtas

9、ks，awoodpropertyanalysissoftwareisimplementedthroughthecombinationofdevelopingenvironmentVisualCwithcomputingtoolMATLABEfficientcommunicationisrealizedthroughembeddingapplicationinterfaceinfriendlyuserinterfaceTheconclus

10、ionofthesisisgiven，andfuturewoodresearchissuesbasedonNNareproposedatthelastKeywords：ChineseFir，72Yang，microstmctare，chemicalcomponent，physicalandmechanicalqualityGeneralizedRegressionNeuralNetwork(GRNN)，applicationinterf