摘要
为了提高单通道盲源分离性能,首先由单路信号利用经验模态分解得到一系列本征模函数分量组合成多路信号;其次针对存在模态混叠的本征模函数分量,提出利用信号周期性构造其多路信号,并利用独立分量分析消除模态混叠的有效方法;然后利用互相关性消除上述步骤所得到的多路信号中的虚假分量,并将剩余的分量信号与观测信号构成新的多路信号;最后利用Fast-ICA(fast-independent component analysis)算法分离得到源信号。仿真实验表明该算法能够有效分离源信号,分离性能优于目前已有的基于经验模态分解的单通道盲源分离算法。
In order to improve the performance of single channel blind source separation, this paper firstly applied the empirical mode decomposition to single-channel observation signal to obtain a series of intrinsic mode function and residue, which reconstructed multi-channel signals. Secondly, according to the existing mixed modes intrinsic mode function, it constructed the multi-channel signals by using the signal periodicity, and eliminated the modal mixing by using independent component analysis. And it cancelled out the false components of the above obtained multi-channel signals on the basis of correlation, then got the new multi-channel signals by the remaining signals and observed signal. Finally, it separated the source signals by using the Fast-ICA algorithm. The simulation results show that the algorithm can effectively separate the source signals, the separation performance is better than the existing single channel blind source separation algorithms based on empirical mode decomposition.
出处
《计算机应用研究》
CSCD
北大核心
2017年第10期3010-3012,共3页
Application Research of Computers
关键词
单通道盲源分离
独立分量分析
经验模态分解
本征模函数
模态混叠
single-channel blind source separation （SCBSS）
independent component analysis （/CA）
empirical mode decomposition （EMD）
intrinsic mode function（IMF）
mixed mode
