【摘要】 针对脑机接口研究(BCI)中对脑电波信号的分类识别问题,对脑电信号中P300脑电信号的预处理、特征提取及特征分类等方面算法进行了研究,主要侧重于对P300脑电信号分类算法的研究。提出了一种自适应的集成支持向量机(SVM)分类方法,利用免疫算法的多样性以及自我调节能力,对基于Bagging的集成SVM分类学习器进行了优化,提高了对P300脑电信号识别的准确度以及针对不同个体的自适应性。研究结果表明:将自适应集成分类算法运用在BCI Competition III Dataset II的P300脑电数据上,可以识别出被试者的脑电意图,并且对P300脑电信号的分类可以达到较高的分类准确率,实验结果稳定在98%。更多还原

【Abstract】 Aiming at the classification and recognition of brain wave signals in brain-computer interaction( BCI),the algorithms of preprocessing,feature extraction and feature classification of P300 signals in brain electrical signals were mainly studied in the research of P300 classification algorithm of EEG signal,an adaptive SVM classification method was proposed to optimize the integrated SVM classification learner based on Bagging by using the diversity of immune algorithm and self-adjusting ability. The accuracy of P300 EEG recognition and self-adaptability to different individuals can be improved by this method. The results indicate that using adaptive integrated classification algorithm on P300 EEG data of BCI Competition III Dataset II can identify subjects’ EEG intent,and classification of P300 EEG can achieve higher classification accuracy. The final experimental results are stable at 98% of the classification accuracy.更多还原