基于集合经验模态分解的脑电信号高阶张量特征提取

为了实现脑机接口系统需要有效的特征提取算法。针对二维主成分分析(2DPCA)的特征提取方法忽略脑电信号(EEG)频域特征的缺点和基于小波分解构建EEG高阶张量时小波参数难以确定的局限性,提出了基于集合经验模态分解(EEMD)构建高阶张量结合多线性主成分分析(MPCA)降维的特征提取方法。设计了3种不同特征提取方法的对照实验,并结合Fisher线性判别分析分类方法取得分类准确率。结果表明:新提出的方法相比基于小波分解构建高阶张量结合MPCA进行降维和2DPCA的特征提取方法,平均识别准确率分别提高4.75%和2.6%,且识别准确率的方差分别减小72.69%和23.86%。该方法在提高单次运动想象脑电信号识别准确率的同时还具有更好的适用性,为实现运动想象脑电信号解码奠定了基础。

Feature Extraction of EEG High Order Tensor Based on EEMD

In view of the shortcomings of 2D principal component analysis (2DPCA)which ignores the frequency domain characteristics of electroencephalography (EEG) and the limitation that wavelet parameters are difficult to determine when constructing EEG high-order tensor based on wavelet decomposition, a feature extraction method based on ensemble empirical mode decomposition (EEMD)and multi linear principal component analysis (MPCA) is proposed. The contrast experiments of three different feature extraction methods are designed, and the classification accuracy is obtained by combining Fisher linear discriminant analysis classification method. The results show that compared with the feature extraction method of constructing high-order tensor based on wavelet decomposition combined with MPCA for dimensionality reduction and 2DPCA, the average recognition accuracy is improved by 4.75% and 2.6% respectively, and the variance of recognition accuracy is reduced by 72.69% and 23.86% respectively. The new feature extraction method not only improves the recognition accuracy of single motor imagery EEG signal, but also has better applicability, which lays the foundation for the realization of motor imagery EEG signal decoding.