基于多类运动想象任务的EEG信号分类研究

针对多类运动想象脑电信号个体差异性强和分类正确率比较低的问题,提出了一种时-空-频域相结合的脑电信号分析方法:首先利用小波包对EEG原始信号进行分解,根据EEG信号的频域分布提取出运动想象脑电节律,通过“一对多”共空间模式(CSP)算法对不同运动想象任务的脑电节律进行空间滤波提取特征;然后将特征向量输入到“一对多”模式下的支持向量机(SVM)中,并利用判断决策函数值的方法对SVM的输出结果进行融合;最后通过引入时间窗对脑电信号进行时域滤波,消除运动想象开始和结束时脑电的波动,进一步提高信号信噪比和算法的分类效果;实验结果显示:在时间窗为2 s时,平均最大Kappa系数达到了0.72,比脑机接口竞赛第一名提高了0.15,验证了该算法能够有效减小脑电信号个体差异性影响,提高多类识别正确率。

Research on EEG Signal Classification Based on Multi - class Motion Imagination Task

For the problem of the individual difference and the classification accuracy of multi class motor imagery EEG signal, a new analysis method for EEG signal based on time-space- frequency domain is put forward:firstly, the wavelet packet is used to decompose the original signal of EEG, and the motor imagery EEG rhythm is extracted according to the frequency distribution of EEG signal, and the spatial features of EEG are extracted from different motor imagery tasks through the "one-to-rest" common space pattern (CSP) algorithm; then the feature vector is input to the support vector machine (SVM) in "one-to-rest" mode, the output value of SVM is fused via the method of judging the decision function value; finally, the time domain window is used to filter the EEG signals to eliminate the fluctuations of the brain at the beginning and end of motor imagery, and further improve the signal to noise ratio and the classification accuracy of the algorithm. The experimental results show that, when the time window is 2 s, the average maximum coefficient is 0.72, which is 0.15 higher than the first of BCI competition. Meanwhile, the results verify that the algorithm can effectively reduce the influences of the individual differences of EEG signals, and improve the accuracy of multi-class recognition.