论文部分内容阅读
分形差分高斯噪声过程是分形布朗运动的一种离散等价情形，它的滤波形式非常适合于描述具有不同的短期间和长期间相关特性的信号。本文研究了滤波分形差分高斯噪声过程的结构辨识问题，针对存在较弱和适中的长期间相关的信号，提出了一种基于超定辅助变量乘积矩和最小描述长度准则的ＯＩＶＰＭ－ＭＤＬ结构辨识方法；针对存在强长期间相关的过程，提出了分形反滤波与ＯＩＶＰＭ－ＭＤＬ的组合定阶方法。方法的可行性与有效性通过大量的数值仿真试验得到验证
The fractal difference Gaussian noise process is a discrete equivalent of fractal Brownian motion. Its filtering form is well suited for describing signals with different short- and long-term correlation properties. In this paper, the structural identification of filter fractal Gaussian noise is studied. For weak and moderate long-term correlation signals, an OIVPM-MDL structure identification method based on product moment and minimum description length of overdetermined auxiliary variables is proposed Aiming at the process of the existence of the strong period, a combination of fractal inverse filtering and OIVPM-MDL is proposed. The feasibility and effectiveness of the method are verified by a large number of numerical simulations