Feature extraction of multiple motor imagery EEG signals is one of the key techniques of BCI. Many studies show that the brain is a nonlinear dynamic system. This paper uses Lyapunov exponent, correlation dimension and approximate entropy of nonlinear dynamics to carry out the feature extraction of eight kinds of motor imagery EEG. The extraction effect is analyzed. It is obtained that the approximate entropy algorithm has the best effect on feature extraction of EEG signals. Finally, the optimal parameters of the approximate entropy algorithm are determined, which provides the basis for the feature extraction of EEG signals.%对多类运动想象的脑电信号进行特征提取是脑机接口的关键技术之一，很多研究表明大脑是一个非线性动力学系统[1]。本文采用非线性动力学的Lyapunov指数、相关维数、近似熵对八类运动想象脑电信号进行特征提取，并分析了其提取效果，得出近似熵算法对脑电信号进行特征提取时效果最好。最后确定了近似熵算法最优参数，为之后的脑电信号特征提取工作提供了依据。
展开▼
