一种改进的强跟踪UKF及其在目标跟踪中的应用

An improved strong tracking UKF and its application in target tracking

针对无迹卡尔曼滤波(UKF)算法缺乏对系统模型和噪声不准确等情况时的自适应调整能力导致滤波精度下降甚至发散的问题,提出了一种改进的强跟踪UKF滤波算法(IST-UKF).首先,结合UKF算法以及强跟踪滤波原理,阐明了强跟踪UKF成立的充分条件.其次,在此基础上,提出在向前一步预测协方差矩阵中引入两个多重次自适应因子,并分别设计了其计算方法.最后,将该算法应用于目标跟踪中,并与强跟踪UKF算法(ST-UKF)以及UKF算法进行仿真对比.仿真结果表明,IST-UKF算法不仅具有强跟踪能力,还能对过程噪声进行自适应调整,实现了对目标的良好跟踪;并且当初始过程噪声设置较大时,更有利于IST-UKF算法的发挥.

Unscented Kalman Filter(UKF) algorithm lacks the adaptive adjustment ability to the system model and noise inaccuracy, which leads to the filtering accuracy decline and even divergence. In order to solve this problem, the paper proposes an improved strong tracking UKF filtering algorithm(IST-UKF). Firstly, combined with the UKF algorithm and the principle of strong tracking filtering, the paper illustrates the sufficient conditions for the establishment of the strong tracking UKF. Then, on this basis, the paper presents the idea of introducing double multiple adaptive factors into the forward prediction covariance matrix, and also designs their calculation methods respectively. Finally, the algorithm is applied to target tracking, and compared with strong tracking UKF algorithm(ST-UKF) and UKF algorithm through simulation. The simulation results show that the IST-UKF algorithm not only has strong tracking ability, but also can adaptively adjust the process noise to achieve good tracking of the target, and that when the initial process noise setting is large, IST-UKF algorithm is more favorable to play.