摘要
脑磁信号作为一种新的脑机接口(BCI)输入信号,含有手运动方向的模式信息。在研究了适用于非平稳性的自适应自回归模型和适用于非高斯性的高阶谱自回归模型的基础上,本文针对脑磁信号的非平稳非高斯性,提出了一种新的特征提取算法,即基于经验模态分解的自回归模型。实验结果表明该算法适合于分析非高斯、非平稳的脑磁信号,结果优于上述的两种算法,并且超过了脑机接口竞赛四优胜者的识别率。
The Magnetoencephalography (MEG) can be used as a control signal for brain computer interface (BCI), which contains the pattern information of the hand movement direction. For non-stationary performance of MEG, an algorithm of the Adaptive Autoregressive (AAR) model is proposed. For non-Gaussian performance of MEG, an algorithm of the Higher-Order Spectral (HOS) and Autoregressive (AR) model is proposed. Based on these two algorithms, we propose an algorithm of the empirical mode decomposition (EMD) -based AR model. Ex- periment results show that the classification accuracy obtained from the EMD-based AR model is higher than those from the other two methods. Furthermore, the proposed method has a higher recognition rate than that of the winner of the 2008 competition Ⅳ.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2010年第7期1460-1465,共6页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金项目(60504035
60605006
60904100)
燕山大学博士基金项目B433
国家大学生创新实验计划项目
关键词
脑机接口
脑磁图
自回归模型
经验模态分解
Brain Computer Interface (BCI)
magnetoencephalography
autoregressive
Empirical Mode Decomposition (EMD)