一种自适应单入多出盲源分离方法

A SELF-ADAPTIVE ALGORITHM FOR SINGLE INPUT MULTIPLE OUT BLIND SOURCE SEPARATION

单入多出盲源分离SIMO_BSS(Single input multiple out blind source separation)是一种特殊的欠定盲源分离情况。目前算法过程中过多地需要根据个人经验判断,自适应差。针对此问题提出通过经验模态分解EEMD(Ensemble empirical mode decomposition)将由多路信号混合成的单路信号分解为多路本征模态函数IMFs(Intrinsic mode functions),采用主成分分析PCA(Principle com-ponent analysis)对多路本征模态函数自适应降维,并利用独立成分分析ICA(Independent component analysis)恢复出相互独立的多路源信号。最后,分别对周期混合信号、生物混合信号进行仿真,仿真结果表明在不同NSR条件下,与EEMD_ICA算法相比,速度快且分离效果较好。

Single input multiple out blind source separation （SIM0_BSS） is a special kind of underdetermined blind source separation. In the process of existing algorithms, too much personal experience is required in judging, and the self-adaptability is poor. To address this problem, we propose that first the ensemble empirical mode decomposition （EEMD） is used to decompose the single-channel signal which is the mixture of muhi-channel signals into multi-channel intrinsic mode functions （IMSs） ; then, the principal component analysis is applied to reduce the dimensionality of multi-channel IMFs adaptively, meanwhile the independent component analysis is used to restore the mutual- independent multiple source signals. Finally, the simulation is conducted on periodic mixed signal and biological mixed signal, the simulation results indicate that the proposed algorithm has quick speed with good separation effect than the EEMD ICA algorithm under different NSR conditions.