EMD和FCM的脑电信号处理方法  

EEG Signal Processing Method Based on EMD and FCM

作　　者：余炜[1] 韩强[1] 马晶晶[1] 谢培[1] 

机构地区：[1]成都工业学院电子工程学院,四川成都611730

出　　处：《数学的实践与认识》2016年第15期223-228,共6页Mathematics in Practice and Theory

基　　金：四川省教育厅重点项目(14ZA0286);四川省科技支撑项目(2013SZZ026)

摘　　要：基于脑电信号非平稳、复杂、信噪比低的特性,使用经验模式分解(EMD)算法对脑电信号进行分解,提取主要IMF分量的特征值,之后使用模糊C-均值(FCM)进行分类,并与现有的几种脑电分类方法做了对比研究.研究结果表明,基于2003年第二届BCI大赛脑电信号库的分类正确率达到78%,对于现有的脑电分类方法有一定的借鉴意义.As EEG is non-stationary,complex,and usually has low signal-to-noise ratio,we used empirical mode decomposition（EMD） algorithm to decompose the EEG and extracted the characteristic values of the major IMF components.Then the Fuzzy C-means clustering（FCM） was used for classification and the results were compared with several present methods.The results showed that the classification accuracy of the proposed method was up to78%based on the database of the second session of 2003 BCI competition.Thus the method could be used as a reference for the present EEG classification methods.

关 键 词：脑-机接口 经验模式分解 模糊聚类 

分 类 号：TN911.7[电子电信—通信与信息系统]