双层无迹卡尔曼滤波

Double Layer Unscented Kalman Filter

针对无迹卡尔曼滤波(Unscented Kalman fllter, UKF)在强非线性系统中估计效果差的问题,提出了双层无迹卡尔曼滤波(Double layer unscented Kalman filter, DLUKF)算法,该算法用带权值的采样点表征先验分布,而后用内层UKF算法对每个采样点进行更新,最后引入外层UKF算法的更新机制得到估计值和估计协方差.仿真结果表明,相比于传统算法,所提的DLUKF算法可以在较低计算负载下获得较高滤波估计精度.

The unscented Kalman filter(UKF)has the problem of the inaccurate estimation in strong nonlinear systems.To solve this problem,the double layer unscented Kalman filter(DLUKF)algorithm is proposed.In the proposed algorithm,the weighted sampling points are used to represent the prior distribution,and then the inner layer UKF algorithm is used to update each sampling point.Finally,the state estimations are obtained by the update mechanism of the outer layer UKF algorithm.Simulation results show that the proposed algorithm not only has a low computational complexity,but also has a very good estimation accuracy,compared with the existing filtering algorithms.