基于HHT的脑电信号在不同阅读模式下的识别与分类

目的：针对Powerlab脑电信号记录仪获取脑电波形，分辨不同类型的脑电阅读模式。方法：对实验者在阅读不同材料（平静闭目、阅读英语、阅读诗歌、阅读现代文四种阅读模式）时的头皮脑电信号进行采样，使用希尔伯特-黄变换及支持向量机训练，分辨平静闭目和其他三种不同阅读模式，并针对经验模态分解时出现的常见情况——端点飞翼现象进行算法优化处理并比较其处理效果。结果：基于多项式拟合处理的经验模态分解分解后的脑电信号辨识率最高，稳定在65%水平，最高可达70%。结论：大脑在阅读状态下经经验模态分解和多项式拟合后的信号适合作为大脑阅读模式下的特征提取函数，并对有效阅读模式具有指导意义。

Recognition and classification of EEG signal in reading mode based on Hilbert-Huang Transformation

Objective To distinguish different kinds of EEG signals from the high-dimensional and redundant mass EEG nonlinear-data by Powerlab. Methods Firstly,EEG signals were sampled from an experimenter’s scalp when the experimenter was reading different kinds of books(closing eyes,reading English books,reading poems and read?ing modern Chinese). Secondly,HHT transform(Hilbert-Huang Transform,HHT)and Support Vector Machine method were used to train and distinguish the model of closing eyes and other three kinds of reading patterns. Final?ly,the algorithm is optimized because of its frequent phenomenon-end issue that occurred during the Empirical Mode Decomposition and the results were analyzed. Results the Empirical Mode Decomposition based on polynomi?al fitting algorithm could be used to recognize largest amount of EEG signals by 70%. Conclusions The experimental results demonstrate that the Optimized HHT algorithm based on Empirical Mode Decomposition and polynomial fit?ting algorithm can effectively make use of the information from the mass EEG nonlinear-data signal and is suitable and practical method of classification for research.