分层向量自回归的多通道脑电信号的特征提取研究

发布时间：2018-01-04 13:00

本文关键词：分层向量自回归的多通道脑电信号的特征提取研究　出处：《自动化学报》2016年08期 　论文类型：期刊论文

【摘要】：有效的特征提取方法能提高脑机接口(Brain-computer interface,BCI)系统对脑电(Electroencephalogram,EEG)信号的识别率.因脑电信号都是多通道的,本文将分层向量自回归(Hierarchical vector autoregression,HVAR)模型用于脑电信号的特征提取,并结合传统的线性支持向量机(Support vector machine,SVM)用于脑电信号识别.该模型不仅克服了自回归(Autoregression,AR)模型只能用来提取单通道特征的局限性,而且不再采用传统VAR(Vector autoregression)模型所有通道共用一个时滞的处理方法.创新之处在于在传统的VAR模型基础上添加正则化思想,有效地压缩参数空间,实现合理的分层结构.本文首次将HVAR模型用于由Keirn等采集并整理的脑电数据中.实验结果证明HVAR模型在阶数较小的情况下(2阶)与阶数较大(6阶)的AR模型效果相当,可见低阶的HVAR能很好地刻画脑电信号的时空关联关系,这说明HVAR可能是刻画EEG信号的一种新颖的方法,这对其他多通道时间序列分析都有借鉴意义.
[Abstract]:Effective feature extraction method can improve the brain computer interface (Brain-computer, interface, BCI) on brain electrical system (Electroencephalogram, EEG) signal recognition rate. Because EEG signals are multi-channel and the hierarchical vector autoregression (Hierarchical vector autoregression, HVAR) model for feature extraction of EEG signal, and combined with the the traditional linear support vector machine (Support vector machine, SVM) for EEG signal recognition. The model not only overcomes the autoregressive (Autoregression, AR) model can only be used to extract the characteristics of single channel limitation, and not by the traditional VAR (Vector Autoregression) method to all channels share a innovation delay. Lies in adding regularization theory based on the traditional VAR model, effectively reduce the parameter space, hierarchical structure to achieve reasonable. This is the first HVAR model will be used by Keirn The EEG data acquisition and consolidation. The experimental results show that the HVAR model in the case of small order (2 order) and the order is large (6 order) equivalent AR model, spatial relationship of visible low order HVAR can well describe the EEG signal, indicating that HVAR may be a a novel method of describing EEG signal, which is of great significance to the analysis of time series of other channels.

【作者单位】：燕山大学信息科学与工程学院;
【基金】：国家自然科学基金(61473339) 中国博士后科学基金(2014M561202) 河北省博士后专项基金(B2014010005) 首批“河北省青年拔尖人才”项目([2013]17)资助~~
【分类号】：R318;TN911.7
【正文快照】： 通过建立计算机与脑电波之间的联系从而实现大脑与外界信息的交换已经成为多年的一个研究热点[1-2].脑机接口(Brain-computer interface,BCI)是一种新的人机交互方式,只需要通过大脑的活动信号将命令传给计算机或其他外部设备[3].这种不依赖外围神经肌肉的通信系统为完全瘫

【相似文献】