【作者】 李丽君； 黄思娟； 吴效明； 熊冬生；

【Author】 LI Li-jun,HUANG Si-juan,WU Xiao-ming,XIONG Dong-sheng(Department of Biomedical Engineering,South China University of Technology,Guangzhou,510006)

【机构】 华南理工大学生物科学与工程学院；

【摘要】 基于脑电信号的事件相关去同步(ERD)和事件相关同步(ERS)这一生理特点,对想象左、右手运动的训练集脑电信号,在频域范围内,采用AR模型功率谱估计法所得C3、C4通道的功率谱密度,确定ERD/ERS较明显的频率范围;在时域范围内,比较C3、C4通道信号的能量差异,确定ERD/ERS较明显的时间段。采用带通滤波和小波包分析的方法提取想象运动过程中ERD/ERS生理现象较明显的节律信号。对比线性分类器,支持向量机(SVM)实现测试集运动想象脑电数据的分类有更好的效果。实验结果表明,能够较准确地提取脑电信号的本质特征,具有较好的抗干扰能力和分类性能,正确率为87.14%。更多还原

【Abstract】 According to event-related desynchronization(ERD)and event-related synchronization(ERS)in imaginary movement,the training set of imaginary left and right movement of the EEG,in the frequency range,using AR model spectrum estimation method to obtain the power spectral density from C3,C4 channel to determine the observably ERD /ERS frequency range;in the time domain,comparing the energy difference between C3 and C4 channel to determine the observably ERD /ERS time section.Use band-pass filter and wavelet packet analysis to extract the observably sigma rhythm of ERD /ERS in imaginary movement.Comparison of linearity distinction analysis,support vector machine(SVM)have better results in classification motor imagery of test set,The results show the method can more accurately extract the essential characteristics of EEG and has good anti-jamming capability and classification performance,the correct rate of 87.14%.更多还原