基于廣義動態(tài)模糊神經(jīng)網(wǎng)絡(luò)的肌電信號模式識別系統(tǒng).pdf

1、華中科技大學(xué)碩士學(xué)位論文基于廣義動態(tài)模糊神經(jīng)網(wǎng)絡(luò)的肌電信號模式識別系統(tǒng)姓名：熊漢申請學(xué)位級別：碩士專業(yè)：控制工程指導(dǎo)教師：黃劍20110528II華中科技大學(xué)碩士學(xué)位論文AbstractInmodernrehabilitationmedicinethereisanincreasingdemofmedicalsystemwiththefunctionofintelligentrehabilitationitisoneoftheindisp

2、ensableproceduretocreatethesystembyrecognizingthesptsconsciousnessofpeopleaccuratelybytheelectromyographic(EMG).ThecontentofthethesisincludestheresearchofacteristicsofEMGhowtorecdEMGhowtoextractthefeatureshowtorecognizet

3、heaction.AccdingtothecompareofseveralalgithmsoffeatureextractionwecombineitwiththepatternrecognitionalgithmstoestablishastableexperimentalsystemcertificatetheeffectofthissystemusingactualEMGsignalsfinallyrecognizethemult

4、iactionmodesofh.AtthebeginningthepaperintroducesthecurrentprogressintheresearchofEMGpatternrecognitionitsrelatedfields.ThenwetakeasystematicapproachtoseveralalgithmsofEMGfeatureextractiontheirsignificanceisgiven.Aftertha

5、tthepapersummarizesthestructuralfeaturesmechanismofthefuzzysystemneuralwkintroducingtheweightoperationsmembershipfunctionionbasicstructureoffuzzyneuralwk(FNN)whichincpatingtheirmeritsawkstructureofaFNNwithdynamicadjustme

6、ntfeature(DFNN)isemphasized.ThenthestructureoftheEMGcollectingdeviceincludingthehardwaresoftwareisintroduced.Afterwardsthepaperdescribesthestardsimplementationprocedureofthealgithmsofpatternrecognitionbasedongeneralizedd

7、ynamicfuzzyneuralwk(GDFNN)indetail.AtlastthispaperprovidesananalysiscomparisoninsevenEMGfeatureextractionalgithmssummarizestheirresultsofpatternrecognition.Theexperimentsindicatethattheaccuracycouldbe97%byusingalgithmsof

8、GDFNNtorecognizeseventypeofactionsithasawideapplicabilityinseveralapproachesoffeatureextraction.Bythedevelopmentofthisexperimentalplatfmweproposedimplementingapproachesofh’smultiactionpatternrecognitionbasedonEMGsignals.

9、InthefuturewewillsearchamesuitablemethodofEMGcollectingstudyingthenewalgithmsoffeatureextraction.Atthesametimewewilladjustthealgithmsofpatternrecognitiontoimprovetherealtimeperfmanceofthesystemlayingasolidfoundationfmoti