快速强跟踪UKF算法及其在机动目标跟踪中的应用

当系统模型不能正确描述真实系统时,强跟踪无迹卡尔曼滤波(unscented Kalman filter, UKF)能很好地弥补传统UKF鲁棒性差的不足,保证滤波精度,但需要额外使用无迹变换,极大地增加计算量。针对这一问题,利用Taylor展开分析渐消因子在UKF中的机理,建立渐消因子近似引入方法,提出快速强跟踪UKF。基于统计浮点运算次数的方法定性分析计算量,表明快速强跟踪UKF计算量与传统UKF相近。根据滤波收敛性判据,讨论了强跟踪UKF的收敛性。仿真实例证明,快速强跟踪UKF滤波精度与强跟踪UKF相差无几,计算量大幅降低。

Speedy strong tracking unscented Kalman filter and its application in maneuvering target tracking

When the system model can not correctly describe the real system, the strong tracking unscented Kalman filter (UKF) can well make up the lack of robustness in the traditional UKF and ensure the accuracy of filtering. However, the computational load of strong tracking UKF is greatly increased due to the additional use of unscented transform. To solve this problem, the Taylor expansion is employed to analyze the mechanism of fading factor in UKF, an approximation fading factor introducing method is established, and the speedy strong tracking UKF is proposed based on the approximation introducing method. A qualitative analysis is carried out using statistical floating point operations (flops)， which shows that the computational load of speedy strong tracking UKF is close to that of traditional UKF. The convergence is discussed based on the filtering convergence criterion. Simulation results demonstrate that speedy strong tracking UKF performs similarly compared with strong tracking UKF, while the computational load has been significantly degraded.