基于泛化高斯模型infomax算法的水声信号盲分离

在信号盲分离领域中,信息最大化算法是一种比较成熟的算法,尤其在处理语音信号盲分离问题中,有着较好的效果.通过对水声信号幅值分布的研究,将基于概率密度函数估计的信息最大化算法运用到水声信号盲分离中,并通过仿真实验比较了基于不同概率密度函数的信息最大化算法对水声信号盲分离的效果.ICA算法的成功与否取决于它的概率密度模型是否能较好的拟合信号本身固有的统计分布.通过对船舶辐射噪声信号的盲分离实验证明了GGM能较好的拟合船舶辐射信号及海洋噪声信号这两种不同的概率分布.

Infomax Based on GGM of the Acoustic Signal Blind Separation

In the domain of the blind separation of signals, the infomax algorithm is a mature arithmetic. Which can achieve a successful separation results, especially in separating the speech signals. By using the arithmetic of infomax into the separation of underwater acoustic signals by researching the underwater acoustic signals distributing. And compare the effects of separating the underwater acoustic signals based on different probability density by doing the experimentations. Whether the arithmetic of ICA is successful or not depends on its probability density model to better fit the inherent statistical signal distribution. The ship radiated noise signals in the blind separation experiments proved GGM can better fit the ship radioactive signals and the marine noise signals these two different probability distribution.