后非线性混合盲信号分离的一种新算法

本文提出了一种新的后非线性混合盲信号分离算法.现存算法大多需要额外的附加源信号信息,才能实现信号的分离,使盲分离变成了半盲分离.鉴于此,本文提出了一种不需要任何附加信息的全盲分离算法.首先,通过微分变换将后非线性混合模型变换成形式如同线性瞬时混合模型的形式,并论证了源信号的微分形式保留了源信号的统计特征.这样,就使非线性问题得到大大简化.其次,利用信号的相火特性建立目标函数及递推方式,用LMS算法使目标函数达到最小值,从而实现了盲信号分离的目的.最后,通过计算机仿真试验验证了本文算法的可行性和有效性.与现存算法相比,本文算法计算量小,收敛速度快,实时性好,实现了全盲分离.

A New Algorithmfor Blind Separation of Post-Nonlinear Mixture

Anewalgorithmfor blind separation of post-nonlinear mixture is proposed inthis paper .The extra information on sourcesignals is required and blindseparation becomes half-blindseparationforthe post-nonlinear mixtureinthe most ofthe existing algo-rithms .So whole-blindseparationalgorithmwithout any extrainformationis proposed.Firstly ,the model of post-nonlinear mixtureistransformedintothe model like the instantaneous linear mixture bythe differential transformation,which makes the nonlinear prob-lemsimplifie...