论文部分内容阅读
将经典的多信号分类算法 (MUSIC)用于研究脑电逆问题时存在两个问题 :对有色噪音敏感和不能识别相干源 .近年人们提出了利用延时相关、高阶累积量或假设已知噪音协方差来缓解有色噪音对算法的影响 .对于相干源 ,则有人提出了递归的多维MUSIC方法 .本文在这些工作的基础上建立了一种基于延时相关阵的、叠代的多维MU SIC算法 .仿真数据及实际脑电应用研究表明 ,该方法能在压制有色噪音的同时识别多个相干源 ,因而具有明显的意义 .
There are two problems when using the classical MUSIC algorithm to study the inverse problem of EEG: it is sensitive to colored noise and can not identify coherent sources. In recent years, it has been proposed that using MUSIC, which is known by using delayed correlation, higher order cumulants or assumptions, Noise covariance to mitigate the impact of colored noise on the algorithm.For the coherent source, then proposed a recursive multi-dimensional MUSIC method.In this paper, based on these work to establish a delay-dependent matrix, iterative multi-dimensional MU SIC The simulation data and practical EEG applications show that this method can identify multiple coherent sources while suppressing colored noise, which has obvious significance.