基于URD的非线性滤波算法

Novel Nonlinear Filter Algorithm Based on URD

非线性滤波算法一直是个活跃的研究方向,本文引入数论中的均匀随机设计(URD),提出了一种URD的非线性滤波算法(URD-KF),用于处理非线性高斯系统的滤波问题,并从理论上分析了该算法的性能。分析表明,该算法的性能要好于Unscented变换方法。仿真结果也证实了该算法的滤波性能优于UK滤波算法,且在相同滤波性能的前提下,该算法的计算代价要小于PF滤波算法。将该算法应用于Bearing-Only跟踪中,实验结果表明,该算法的性能优于PF。

The research on nonlinear filter algorithm is an active orientation. Based on the uniformly random design （URD） in number theory, the nonlinear filter algorithm is presented. A novel approach for filtering nonlinear Ganssian system is made. The URD-KF is introduced. The analysis shows that the mean and covariance predicted by the new methods are more accurate than unscented transform. The simulation results also show that the performance of the algorithm is better than UK. The application to the Bearing-Only Tracking also illustrates that it outperform the PF.