针对在单站无源定位中UKF算法由于数值计算的舍入误差带来的滤波发散以及计算量较大问题,文中提出了一种 采用奇异值分解和超球体采样的UKF滤波算法。该算法将标准UKF算法中的协方差矩阵进行奇异值分解,避免算法在 递推过程中,由于计算舍入误差而引起协方差矩阵失去正定性,而导致算法失效的问题。并且新算法采用超球体采样策 略,减小了采样点的数量,提高了计算效率。仿真实验结果证明了该算法的有效性。

As the calculation of unscented Kalman filter is large and will divergent because of the numerical calculation error, an improved spherical simplex sampling UKF algorithm based on the singular value decomposition (SVD) is presented. To avoiding the invalidation caused by errors during computation, the algorithm uses SVD technique to decompose the covariance matrix. To improve the computational efficiency, the algorithm uses spherical simplex sampling strategy to reduce the number of sampling points. Simulation results show that the validity of the proposed algorithm further.

国家自然科学基金资助项目