基于小波包分解与近似熵的脑电特征提取方法研究及在脑机接口中的应用

发布时间：2018-03-10 21:13

  本文选题：脑机接口　切入点：运动想象　出处：《南昌大学学报(理科版)》2017年03期 　论文类型：期刊论文

【摘要】：为提高运动想象脑机接口的分类正确率,结合小波包分解与近似熵对脑电信号进行特征提取。该方法利用小波包对脑电信号全频段进行分解,用近似熵函数对小波包结点提取分类特征,然后用稀疏表示对特征向量进行降维,最后使用功率差方法进行分类。实验结果表明,在使用1秒数据进行分类的条件下,该方法在使用2种不同通道集合时都取得了很好的分类效果。使用32个和10个通道时分类正确率分别达到了95.65%和86.41%,比小波包分解与空域滤波方法分别提高了5.9%和8.32%,比传统的共空域模式方法分别提高了7.18%和7.27%。另外,使用的数据长度越短,分类识别率越高,表明该方法更适用于较短的数据,有利于提高脑机接口的信息传输速度。
[Abstract]:In order to improve the classification accuracy of the motion-imaginary brain-computer interface, the wavelet packet decomposition and approximate entropy are combined to extract the features of the EEG signal, and the wavelet packet is used to decompose the whole frequency band of the EEG signal. The approximate entropy function is used to extract the classification feature of wavelet packet nodes, then the sparse representation is used to reduce the dimension of the feature vector, and the power difference method is used to classify the feature vector. The experimental results show that, under the condition that 1 seconds data is used for classification, The classification accuracy of 32 channels and 10 channels is 95.65% and 86.41 respectively, which is 5.9% and 5.9% higher than that of wavelet packet decomposition and spatial filtering, respectively. 8.32, which is 7.18% and 7.27 higher than the traditional common airspace mode, respectively. The shorter the length of the data used, the higher the classification recognition rate, which indicates that the proposed method is more suitable for shorter data and can improve the speed of BCI information transmission.
【作者单位】： 南昌大学电子信息工程系;
【基金】：国家自然科学基金项目(61365013,61663025) 江西省教育厅科技项目(GJJ13054)
【分类号】：R318;TN911.7
，

本文编号：1595093