非线性非高斯秩滤波方法

基于秩滤波原理,提出一种非线性非高斯秩滤波方法,给出其递推过程.目前常用的非线性滤波方法有无迹Kalman滤波和粒子滤波,无迹Kalman滤波只适用于高斯分布的情况,粒子滤波方法却存在粒子退化及重采样引起的粒子贫化问题.而非线性非高斯秩滤波方法不仅适用于常见的多元t分布、多元极值分布等非高斯分布的非线性滤波,并且计算简单、工作量小,便于工程应用.从仿真算例可以看到,该方法的滤波精度与无迹Kalman滤波和粒子滤波方法相比提高了500%以上.

Nonlinear and non-Guassian rank filter method

A nonlinear and non-Guassian rank filter (RF) method based on the principle of rank filter was presented. Furthermore, its filter recursive process was also given. Unscented Kalman filter(UKF) only for Gaussian distribution and particle filter (PF) for non-Gaussian distribution were the two common nonlinear filter methods, but PF had the problems of particle degeneracy, particle impoverishment caused by resampling and complicated calculation. Compared with the two methods, the proposed RF is suitable for not only Gaussian distribution but also non-Gaussian distributions such as multivariate t distributions and extreme value distributions, and RF is simple to calculate and easy to apply in engineering. Moreover, it has low amount of calculation. From the simulation comparisons of RF, UKF and PF in the example, the filtering accuracy of RF increased at least 500%.