本研究提出采用一种分段的Prony方法(PPM),将非平稳EEG信号分解为幅度为指数上升或下降的正弦波,从而获得各频率成分的幅度、频率和初始相位,并计算两通道相位差;用香侬熵来衡量两路信号的相位同步程度。首先对该方法的频率、相位分辨率和检测同步程度的效果等进行仿真研究,结果显示了该方法具有高的频率和相位分辨率。最后将该方法用于实际的三组脑电信号,得到的结果与经典方法一致。将该方法与希尔伯特变换方法进行了比较,说明该方法具有较好的抗噪性能,可作为非平稳信号的相位同步检测的有效工具。
In this study, a piecewise Prony method (PPM) was adopted to decompose the nonstationary EEG signals into a summation of sinusoidal components with decaying or growing envelope in order to obtain the magnitude of the frequency components, frequency and phase information. Then, the phase information was used to calculate difference of the phase of two-channel signal. The degree of phase synchronization of two-channel signals was measured according to Shannon entropy. The proposed method was firstly evaluated on simulated transient nonstationary data. The results showed that PPM could estimate the actual frequency and phase with high resolution. With application to three real rat EEG data groups, the phase synchronization entropy of PPM was consistent with that of the classical methods. Compared with Hilbert method, PPM based synchronization analysis was more robust to noise and non-stationary.