Application of Improved Grey Theory in Brain Computer Interface

摘要

Since brain is a grey system, the grey theory is suitable for electroencephalogram (EEG)-based brain computer interface (BCI) system. However, the detection accuracy of EEG signal using traditional grey theory is limited. In the current study, in order to increase the distinguishing ability of the GM (1, 1), we defined new parameters by introducing a factor u to revise developing coefficient and grey action. Consequently, the grey theory has been applied to the BCI system, where the improved grey model was proposed to analyze alpha wave to switch the state of BCI system. Moreover, since Steady-State Visual Evoked Potential Signal (SSVEP) is a kind of oscillation signal with specific frequency, the grey relational analysis has been applied in frequency domain to recognize instruction of the subject. According to the result, the average accuracy of alpha wave detection is 92.5% when the timescale is 0.5s, which is 15% higher than the traditional grey model, and average accuracy of SSVEP detection of four patterns is 86.1% in 3s, which showed that brain-computer interface system based on the grey theory can effectively realize the state transition and instruction recognition.