一种基于逼近线性空间的盲信号分离方法

文章结合目标函数，给出了一种非线性混叠信号盲分离算法，含噪混合信号之间的关系构造了动态的目标函数状态一空间。该函数的特点是多变量，过程和观测噪声不限于高斯分布，是解决该问题的有效方法，从而把非线性混叠信号盲分离问题转化为参数空间的线性混叠信号盲分离问题。在目标函数空间中，也可应用多次分离对信号进行分离。该算法收敛精度较高，稳定性好。

Blind Signal Separation based on a Linear Approximation of the Objective

In combination of objective function, and by applying it in the aliasing situation, a nonlinear aliasing signal blind separation algoritbm is given. Tbe dynamic objective function condition-space is constructed by the relations between chirp-contained composite signals. This function features multivariable, the process and the observation noise not limited to the Gaussian distribution, and is the effective method for solving this problem. Thus the aliasing signal blind separation is transformed into the parameter space linear aliasing signal blind separation. In objective function space, the signal separation may be carried on many times. This algorithm is high in convergence precision and good in stability.