一种基于罚函数的盲分离算法

通过对Hyvarinen-Oja带约束项的学习算法的理论分析与仿真比较,提出了一种基于乘子法罚函数的实时线性混叠盲信号分离算法。该算法不仅从理论上导出了罚函数的选取方法,而且也大大加快了算法的收敛速度,避免了目前纯罚函数方法可能带来的病态问题。仿真结果表明该算法不仅具有很好的分离效果,而且分离时间较Hyvarinen-Oja带约束项的学习算法快。

A Blind Separation Algorithm Based on Penalty Function

Based on a theoretical analysis and simulation comparison of learing algorithms of Hyvarinen-Oja,a real-time linear and mixed blind signal separation algorithm is proposed,which is based on multiplier penalty function,not only the selection of penalty function is deduced theoretically,also the convergence speed of the algorithm is improved greatly,by which the ill condition problems that other penalty function may induce are avoided. The simulation result shows that the algorithm has a good separation effect and it is faster than that of Hyvarinen-Oja.