利用EEG信号的小波包变换与非线性分析实现精神疲劳状态的判定

Evaluation of human mental stress states based on wavelet package transformation and nonlinear analysis of EEG signals

EEG(脑电)信号的4个节律(δ波、θ波、α波、β波)与人的精神疲劳状态有对应关系,不同节律的能量值及其非线性特征参数可以用于疲劳状态的判定。本文首先利用小波包分解与重构技术,构造了以"db20"为基小波函数的6层分解,得到EEG信号的4个节律。然后,对4个节律信号分别计算相应的节律的频带能量比例值,这些频带能量比例值作为对人体精神状态进行评价的量化指标。通过计算EEG信号α波的非线性特征参数,包括最大Lyapunov指数、近似熵、复杂度,并将这些非线性特征参数组成疲劳状态的综合评估判据,可以实现疲劳状态的判定。10组EEG信号的分析结果表明了该本文方法的有效性,其中对疲劳和非疲劳状态的判定准确率较高,而对轻微疲劳、中等疲劳和严重疲劳三种状态的准确区分稍差一些。

The positive correlation is known between the four rhythms of human Electroencephalogram （EEG） signals including δ wave, θ wave, αwave and β wave and human mental stress states. So the energy values of the four rhythms of EEG together with their nonlinear parameters can be used to evaluate mental stress states. Here, the four rhythms of EEG was firstly reconstructed by using the technique of wavelet package transformation, where a 6-level-frame was achieved to decompose the original EEG signal with help of the basis wavelet function of ＂ db20＂. Then, the corresponding frequency-band energy ratio （FBER） of each rhythm was calculated and used to estimate states of mental stress quantitatively. Some nonlinear parameters of c~ wave including maximum Lyapunov exponent, approximated entropy and complexity level were also calculated and a synthesized evaluating criterion was made to determine human mental stress states. The proposed method was verified to be effective with 10 sets of EEG data. It was shown that its accuracy is higher when evaluating fatigue or non-fatigue states; meanwhile, it is not so good to identify the different mental stress states of weak, middle and serious fatigues.