一种非白非高斯数据的数值仿真方法

使用复合抽样法,可以产生具有指定概率密度形式的加性分布非高斯序列。通过在极零图上直接指定数对极零点,可以实现定性色化低阶自回归滤波器设计。把非高斯激励序列通过自回归滤波器,即可得到非高斯信号处理仿真研究中频繁使用的非高斯有色序列。结合一组混合高斯有色数据数值仿真实例,演示了这一由复合抽样法加定性色化构成的非白非高斯数据快捷数值仿真方法的有效性。

A simulation approach to colored non-Gaussian processes

By utilizing the composite sampling method, the non-Gaussian sequence with a certain additive distribution form can be generated. By assigning the couples of poles on the zero-pole plot, the low order autoregressive filter used for qualitative colorization can be realized. After driving the non-Gaussian sequence through the autoregressive filter, the colored non-Gaussian process, which is often used in the simulation study of non-Gaussian signal processing, is obtained. A numerical simulation instance of colored Gaussian mixture data is given to illustrate the performance of this approach of combing composite sampling and qualitative colorization.