摘要
为了提高运动想象脑电信号在少数几个通道情况下的分类性能,提出一种基于变分模态分解(VMD)和共空间模式(CSP)的特征提取方法。首先将原始脑电信号进行预处理,然后将脑电信号进行变分模态分解得到固有模态函数(IMF),通过巴氏距离计算原始信号与各IMF分量之间的相似性,选择合适的IMF分量构造新的信号矩阵,利用CSP空域滤波提取特征,最后使用支持向量机(SVM)实现分类。实验结果表明,对BCI2003数据集进行处理后,所提出的方法分类准确率可达91.43%,并与其它算法进行了比较。同时在实测数据中进行了验证,证明了方法的有效性。
In order to improve the classification performance of motor imagery electroencephalograph(EEG)signals in a few channels,a feature extraction method based on Variational Mode Decomposition(VMD)and Common Spatial Pattern(CSP)was proposed.Firstly,the original EEG signal was preprocessed,and VMD was proposed to decompose the EEG signals into the Intrinsic Mode Function(IMF).The Bhattacharyya distance between the original signal and each IMF component was calculated,and the appropriate IMF components were selected to construct a new signal matrix,CSP was used to extract features.Finally,the classification was performed using Support Vector Machine(SVM).Experimental results show that after processing the BCI2003 data set,the classification accuracy of the proposed method can reach 91.43%,and compared with other algorithms.The experiment analysis of the proposed method was carried out using the measured data,which proves the effectiveness of the method.
作者
刘帅
乌日开西·艾依提
LIU Shuai;Wurikaixi·Aiyiti(School of Mechanical Engineering,Xinjiang University,Urumqi Xinjiang 830047,China)
出处
《计算机仿真》
北大核心
2022年第11期432-437,共6页
Computer Simulation
基金
新疆维吾尔自治区天山英才工程(201720045)。
关键词
运动想象
变分模态分解
共空间模式
巴氏距离
Motor imagery
Variational mode decomposition(VMD)
Common spatial pattern(CSP)
Bhattacharyya distance
