宇航学报 ›› 2019, Vol. 40 ›› Issue (2): 215-222.doi: 10.3873/j.issn.1000-1328.2019.02.011

• 制导、导航、控制与电子 • 上一篇    下一篇

改进的平方根UKF在再入滑翔目标跟踪中的应用

叶泽浩,毕红葵,谭贤四,曲智国,张裕禄,程杨   

  1. 空军预警学院,武汉 430019
  • 收稿日期:2018-05-08 修回日期:2018-09-06 出版日期:2019-02-15 发布日期:2019-02-25
  • 基金资助:

    国家自然科学基金(61401504);博士后科学基金(2014M562562)

Improved Square Root UKF Applying to Reentry Glide Target Tracking

YE Ze hao, BI Hong kui, TAN Xian si, QU Zhi guo, ZHANG Yu lu, CHENG Yang   

  1. Air Force Early Warning Academy, Wuhan 430019, China
  • Received:2018-05-08 Revised:2018-09-06 Online:2019-02-15 Published:2019-02-25

摘要:

针对临近空间高超声速再入滑翔目标跟踪问题,提出了一种基于新气动力模型的改进的平方根UKF滤波算法(ISR-UKF)。首先对气动力模型进行了变换。其次,在传统平方根UKF基础上,改用球形无迹变换来计算权系数以及sigma点;改进了平方根矩阵的分解方法;同时为解决矩阵求逆易出现奇异值使滤波失效的问题,提出在协方差矩阵更新中引入多重次稳定因子。最后将该算法分别与基于原气动力的ISR-UKF,基于新气动力的平方根UKF以及基于原气动力的平方根UKF进行仿真比较。仿真表明,该算法具有良好的滤波性能,而且能避免奇异值问题的出现,具有很好的可靠性。

关键词: 临近空间, 跟踪, 平方根UKF, 气动力模型, 球形无迹变换

Abstract:

To improve the tracking problem of the hypersonic reentry glide targets in near space, an improved square root UKF filtering algorithm (ISR-UKF) based on the new aerodynamic model is proposed. Firstly, we improve the aerodynamic model. Secondly, based on the traditional square root UKF algorithm, we use the spherical unscented transform to calculate the weight coefficient and the sigma point. And we improve the decomposition method of the square root matrix. At the same time, to solve the problem that the singular value makes the filter invalid when the matrix inversion occurs, multiple stability factors are introduced in the covariance matrix update. Finally, the algorithm is compared with the ISR-UKF based on the traditional aerodynamic force, square root UKF based on the new aerodynamic force, and square root UKF based on the traditional aerodynamic force. The results show that the algorithm has good filtering performance, and can avoid the occurrence of the singular value problems. So it also has good reliability.

Key words:  Near space, Tracking, Square root UKF, Aerodynamic model, Spherical unscented transform

中图分类号: