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. Based on the characteristics of non-stationary, complex and low signal-to-noise ratio of EEG signals, EEG (EMD) algorithm was used to decompose the EEG signals to extract the eigenvalues ​​of the main IMF components, and then classified by using fuzzy C-means (FCM) , And compared with several existing methods of EEG classification.The results show that based on the 2003 second BCI Competition EEG signal classification accuracy rate of 78%, for the existing EEG classification methods are Certain reference significance.