一种双谱估计的非参数化方法

A Nonparametric Method of Bispectrum Estimation

传统的信号处理中,大都是以二阶统计量来作为分析信号的主要方法。当系统是非线性、非最小相位系统以及系统输入是非高斯型的时候,传统的二阶统计量方法存在较大的局限性,甚至不能有效地处理信号。为了解决上述问题,人们通常采用高阶谱估计。本文讨论了高阶谱估计的详细原理,并重点对高阶谱估计中应用较为广泛的双谱估计进行了深入的探讨,提出了一种双谱估计的非参数化方法。实验结果表明,文中方法要优于传统的二阶功率谱估计。

In traditional signal processing, the main method of analyzing signal is the second order statistic. When the system is a non-linear, non-minimum phase one and the system input is non-Gaussian signal, the traditional second-order statistic method has great limitation, which cannot even process signal effectively. In order to solve these problems, high order spectrum estimation is usually adopted. In this paper, we detail the principle of the higher order spectrum estimation, discuss the widely-applied bispectrum estimation in higher order spectrum estimation, and put forward a double spectrum estimation of parametric method. The experimental results show that the method is superior to the conventional second-order power spectrum estimation.