一种新的混合迭代UKF

A New Kind of Hybrid Iterated Unscented Kalman Filter

从统计线性回归的角度对无味变换(unscented transformation,UT)进行分析,推导了迭代无味卡尔曼滤波(iterated unscented Kalman filter,IUKF)。针对IUKF计算量大的问题,结合弦线迭代法和IUKF,得到了一种新的混合迭代无味卡尔曼滤波器。数值仿真的结果表明,新滤波算法的精度优于扩展卡尔曼滤波、迭代扩展卡尔曼滤波和无味卡尔曼滤波,并可以有效降低IUKF的计算量。

Unscented transformation is analyzed through the viewpoint of statistical linear re- gression, and iterated unscented Kalman filter （IUKF） is derived. The so-called hybrid iter- ated unscented Kalman filter is presented by incorperating secant method into IUKF in order to cope with the high-computation-cost problem. New filtering method is introduced into the example of univariate nonstationary growth model. Simulation results show that new method outperforms extended Kalman filter, iterated extended Kalman filter and unscented Kalman filter. The computation cost can be reduced relative to IUKF.