摘要
针对现有无迹卡尔曼滤波(UKF)算法在高维系统中易出现协方差非正定导致滤波不稳定甚至发散的问题,探讨了基于方差平方根的容积卡尔曼滤波(CKF)算法在鱼雷目标跟踪中的应用。该算法首先基于Cubature准则,获得一组具有相同权重的Cubature点,然后经过非线性系统方程将该点集进行转换产生新的点,以此预测下一时刻系统的状态,并在滤波更新过程中通过传播状态的方差平方根,确保了方差矩阵的对称性和正定性。仿真结果表明,CKF的滤波精度要高于UKF。
The existing unscented Kalman filter（UKF） algorithm results in non-definite covariance easily in high-dimensional system, which leads to instability of filter and even divergence. In this paper, the application of cuba- ture Kalman filter（CKF） algorithm based on square root of variances to underwater target tracking is discussed. This algorithm can achieve a group of cubature points with same weight based on the cubature principle, transform the points set into the new points by non-linear system equation in order to predict the system state at next time, and ensure symmetry and positive definiteness of the covariance matrix by using the square root of the variances of propagation state. Simulation results show that CKF is better than UKF in estimation precision.
出处
《鱼雷技术》
2015年第6期428-432,共5页
Torpedo Technology
关键词
鱼雷
目标跟踪
方差平方根
无迹卡尔曼滤波
容积卡尔曼滤波
torpedo
target tracking
square root of variance
unscented Kalman filter（UKF）
cubature Kalman filter（CKF）
