Author Login Editor-in-Chief Peer Review Editor Work Office Work

Computer Engineering ›› 2007, Vol. 33 ›› Issue (01): 190-192. doi: 10.3969/j.issn.1000-3428.2007.01.066

• Artificial Intelligence and Recognition Technology • Previous Articles     Next Articles

Study on Brain Computer Interface Based on Cognitive Task

BI Luzheng1, ZHANG Ran2, GAO Yuan3, WU Pingdong1   

  1. (1. School of Mechanical and Vehicular Engineering, Beijing Institute of Technology, Bejing 100081; 2. School of Life Science and Technology, Beijing Institute of Technology, Beijing 100081; 3. School of Information Science and Technology, Beijing Institute of Technology, Beijing 100081)
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-01-05 Published:2007-01-05

基于认知任务的脑机接口方法研究

毕路拯1,张 然2,高 原3,吴平东1   

  1. (1. 北京理工大学机械与车辆工程学院,北京 100081;2. 北京理工大学生命与技术学院,北京 100081;
    3. 北京理工大学信息科学技术学院,北京 100081)

Abstract: A brain computer interface presented using electroencephalogram(EEG) signals are the subjects that have to think of the multiplication task. EEG signals from 4 subjects are recorded at occipital scalp, while they are in the state of the multiplication task and the resting state. The spectral power in the 3 bands: 8~10Hz, 11~13Hz and 14~30Hz, is estimated using the Welch method respectively. A ratio of the average of the spectral power in each of the three bands to the average in 2~30Hz is designed as classification features. The multiplication task is detected by a support vector machine classifier. The experimental results show that the method is feasible, practical, and all accuracies are more than 94.44%, while maxim accuracy is 98.89%. Because the two channels are only used, the method is more convenient in practice for constructing letters, controlling a wheelchair and so on.

Key words: Support vector machine, Brain-computer interface, Spectral power, Cognitive task

摘要: 提出一种通过脑电波来识别放松状态以及乘法作业状态从而实现脑机接口的新方法。利用脑电仪记录受测者放松状态以及乘法作业时的大脑左右半球枕叶部的脑电信号,采用Welch法分别估计出这2个部位8Hz~10Hz、1Hz~13Hz、14Hz~30Hz 3个频段的功率谱,以各个功率谱平均值和2Hz~30Hz频段功率谱平均值的比值作为分类特征,采用支持向量机的方法建立了分类器,从而实现了脑机接口。4个受测者的实验结果表明识别准确率都大于94.44%,最高为98.89%。由于只采用了2个采集点,因此如果采用某种编码方式,该脑机接口技术就可更加方便地用于写字、控制轮椅等方面。

关键词: 支持向量机, 脑机接口, 功率谱, 认知任务