3、江省自然科学基金(Y1111189,LY12F03007);浙江省科技计划项目(2012C33075)*通信作者。E-mail:zqz@hdu.edu.cnZHANGQi-ZhongXIXu-GangMAYu-LiangLUOZhi-ZengSHEQing-Shan(InstituteofIntelligentControlandRobotics,HangzhouDianziUniversity,Hangzhou310018,China)Abstract:Action pattern recognitionoflimbsusingsEMGisthebasis for bioniccontrol
5、meantime,a K Nearest Neighbor (KNN)modelincrementallearningmethod with incrementallearningability,was presented as a classifierofpatternrecognition.Inpatternrecognitionexperimentofclassifyingfourfine movements ofthewrist(namelywristextension,wristflexion,wristpronation,wristsupination)with10particip
6、ants,thecorrectmoderecognitionrateisabove92.5%. In a contrastexperimentthat wasdesignedtoevaluatethe effectsof theincrementlearningabilitytothe action mode recognitionrate,thecorrectrecognitionrateis4.5percenthigherthanKNNmodearithmeticwithoutincrementallearningability whentheprostheticusers change