基于改进FastICA的雷达信号盲分离算法

Algorithm of Blind Radar Signal Separation Based on Improved FastICA

在复杂的电磁环境中,对频谱混合的雷达信号进行盲分离(BSS)处理不失为一种有效的信号分选手段。而传统的盲分离算法存在收敛速度慢,对初始值要求较高等缺点,极大地影响了分离效果。在独立分量分析(ICA)模型的基础上,利用最大化的负熵作为目标函数,提出了一种采用阻尼牛顿法与罚函数相结合的改进算法对其进行寻优,将原问题转化为无约束的收敛问题,加快了收敛速度。采用该方法对五路随机混合的雷达信号进行仿真实验,成功得出了分离信号,有效提高了收敛速度,简化了分离条件,达到了较好的盲分离效果。

In the complex electromagnetic environment, mixed spectrum radar signal blind separation is an effective means of signal sorting. The traditional blind separation algorithm convergence slow, require a higher initial value and other shortcomings, greatly influenced the separation. On the basis of （ICA） model on independent component analysis, using of the maximize of the negative entropy as the objective function, raise a new improved algorithm that combine damping Newton method with penalty function to optimize it, change the original problem to a unconstrained problem, increase the conver- gence speed. Using this method to make simulation experiment of five random mixing radar signal, successfully obtain the separation signal, effectively improve the convergence rate, simplify the separation conditions and achieve a better blind sig- nal separation effect.