本研究的目的是發展應用腦電波量測來分析受測者的眼動訊號,透過時域訊號的特徵來辨識眼動狀態,進而了解受測者的疲勞程度。本研究使用由神念科技(NeuroSky)所研發的非侵入式腦電波量測儀,進行腦電波原始資料的擷取,使用Matlab程式對於腦電波進行小波轉換(Wavelet transform, WT)將訊號分解,再將數據轉化為特徵值之後,使用支持向量機(Support vector machine, SVM)與倒傳遞類神經網路(Back propagation neural network, BPNN)進行辨識開閉眼的狀態;透過模糊邏輯,根據閉開眼頻率與閉眼時間來推論受測者之疲勞狀態,進而用於駕駛之警示。
In this study, a drowsiness identification system using mindwave EEG signal is proposed. With the noninvasive mindwave headset developed by NeuroSky, the time domain signal of the mindwave is used to recognize eye movement and the user's fatigue level. First, the EEG raw signal is transformed by the wavelet transformation. Second, the eigenvalues are computed based on the Daubechies wavelet. Third, the support vector machine and the back propagation neural network are studied to identify the status of eye movement using the eigenvalues. Finally, the fuzzy logic is used to obtain the fatigue level, according to the frequency of the eye movement and the time of closing eyes.