论文部分内容阅读
为减少脑机接口的电极数量,采用基于最优化少量电极的共空间模式(CSP)算法提取不同思维任务下的脑电信号(EEG)特征值.首先,根据事件相关(去)同步化现象观察时频特性;然后,运用熵准则对单个电极进行可分性度量;最后,根据可分性排序,利用基于最优化少量电极的CSP算法和支持向量机算法对不同电极组合的特征值进行提取和分类,得出最优化的电极组合.结果表明,进行心算和想像空间旋转2种思维任务时被试的EEG信号在顶叶和枕叶区域存在明显的能量差异,6个被试可分性最高的电极均位于这2个区域;与传统的EEG信号处理算法相比,基于最优化少量电极的算法可以使系统使用的电极数减少至3.3个,并且分类正确率提高5.4%.因此,采用基于最优化少量电极的算法可以减少电极数目,改善思维任务脑机接口的性能.
In order to reduce the number of electrodes in the brain-computer interface, EEG (EEG) feature values under different mental tasks were extracted by using the common space mode (CSP) algorithm based on optimized small electrodes.First, according to the event-related Then the entropy criterion is used to measure the separability of single electrode. Finally, according to the separability sort, the eigenvalues of different electrode combinations are extracted by the CSP algorithm and support vector machine The results show that EEG signals have significant energy differences in the parietal lobe and occipital lobe regions when the two mental tasks of mental arithmetic and imaginal spatial rotation are performed, and the highest divisibility of the six subjects Compared with the traditional EEG signal processing algorithm, the algorithm based on the optimization of a small number of electrodes can reduce the number of electrodes used in the system to 3.3 and improve the classification accuracy by 5.4% Algorithms that optimize a small number of electrodes can reduce the number of electrodes and improve the performance of brain-computer interfaces for thinking tasks.