迭代平方根UKF

Iterated Square Root Unscented Kalman Filter

针对无迹卡尔曼滤波器(UKF)测量更新方法的不足,提出了一种对UKF进行迭代测量更新的方法,用于提高非线性系统状态估计的近似精度.利用平方根UKF算法确保了迭代UKF的数值稳定性.理论分析与实验结果表明,迭代平方根UKF算法不仅具有无需计算雅可比矩阵的优点,而且具有较高的非线性近似精度、较强的数值稳定性和较高的运算效率;在相同数量级运算时间的条件下,其估计性能明显优于扩展卡尔曼滤波器(extended Kalman filter,EKF)、UKF和迭代UKF等非线性滤波器.

In order to improve the standard measurement update procedure of unscented Kalman filter （UKF） and to increase the approximation accuracy of nonlinear state estimates, an iterated measurement update procedure is presented. To guarantee numerical stability of the iterated UKF, a square root version of the UKF is included. Theoretical analysis and simulation results show that the iterated square root UKF （ISR-UKF） does not need to calculate the Jacobian matrix, and with its higher nonlinear approximation accuracy, stronger numerical stability and computational efficiency, the presented ISR-UKF has a better performance than EKF, UKF and the iterated UKF （IUKF） under the same computation burden.