论文标题
关于一步UKF和两步UKF的准确性
On the Accuracy of the One-step UKF and the Two-step UKF
论文作者
论文摘要
最准确的无味卡尔曼过滤器（UKF）的版本涉及建造两个合奏。但是，为了降低计算成本，通常在没有第二集合的情况下实施UKF。但是，这种简化是有代价的，但是，对于线性系统，两步UKF的一步变化并不专门针对经典的卡尔曼过滤器，并且与准确性相关。本文通过开发一个修改后的一步UKF来纠正这种缺点，该UKF恢复了用于线性系统的古典卡尔曼过滤器。数值示例表明，经过修改的一步UKF还恢复了具有线性输出的非线性系统中两步UKF的准确性。
The most accurate version of the unscented Kalman filter (UKF) involves the construction of two ensembles. To reduce computational cost, however, UKF is often implemented without the second ensemble. This simplification comes at a price, however, since, for linear systems, the one-step variation of the two-step UKF does not specialize to the classical Kalman filter, with an associated loss of accuracy. This paper remedies this drawback by developing a modified one-step UKF that recovers the classical Kalman filter for linear systems. Numerical examples show that the modified one-step UKF also recovers the accuracy of the two-step UKF in nonlinear systems with linear outputs.
