改进G-P算法与睡眠脑电的关联维

An Improved G-P Algorithm and Correlation Dimension of Sleep EEG

脑电的非线性参数能有效表达大脑的不同生理状态 ,不同睡眠期脑电的关联维数不同。但用G P算法求关联维数存在抗干扰能力较差、可靠性不稳定、运算量巨大等缺点。先对相空间进行奇异谱分析 ,进而对原始相空间进行旋转 ,使其成为正交的等效空间 ,然后再使用G P算法。改进后的算法能有效地抑制噪声干扰 ,降低相空间规模 ,减少运算复杂性 ,在睡眠脑电的关联维数计算上效果良好。

Non-linear features of EEG represent effectively different physiology states of brain, correlation dimension is different in different sleep stages. But G-P algorithm which be used to calculate D2 have some shortages, such as being robust against noise, not stable, and needing a big operation. An improved approach as follows: Firstly,to apply singular system analysis on phase space, transform the original phase space by circumvolving to a orthogonal equivalent space, then use G-P algorithm. This improved algorithm restrains noise interfere, depresses phase space dimensions, reduce calculation amount, and is favorable in calculating D2 of sleep EEG.