摘要
针对脑电信号非平稳非线性特征,提出基于改进的局部均值分解（Local Mean Decomposition,LMD）运动想象信号分类方法。首先结合改进LMD算法和加窗原则选取4-6 s想象信号作为分类数据,提取包含μ节律和β节律的PF分量;其次计算所选分量的样本熵值;最后用支持向量机进行分类预测,并用分类准确率进行评估。实验结果表明,运用改进LMD比传统LMD方法的识别率更高,从而验证该方法的有效性。
For the non- stationary and non- linear characteristics of electroencephalogram（ EEG）, this paper proposes a classification method based improved local mean decomposition（ LMD） for motor imagery signal. Firstly, combining the improved LMD with window principle to select imagery signal of 4 - 6 second as classification data, and extract components PF that contain μ rhythm and βrhythm. Secondly, the sample entropy of corresponding components PF is calculated. Finally, the EEG is classified with support vector machine（ SVM） and evaluated by the accuracy. The experiment results indicate that the improved LMD algorithm is better than traditional LMD algorithm in classification accuracy, which turns out the effectiveness of proposed approach.
出处
《电子技术应用》
北大核心
2016年第3期116-119,共4页
Application of Electronic Technique
基金
山西省青年基金项目(2013021016-3)
关键词
LMD
加窗原则
样本熵
PF分量
支持向量机
local mean decomposition
window principle
sample entropy
PF components
support vector machine
