组合导航系统滤波器截断误差抑制方法

发布时间：2018-06-15 02:30

本文选题：组合导航 + Bierman-Thorton算法　；参考：《传感技术学报》2014年05期

【摘要】：组合导航系统作为重要的定位和姿态测量的技术手段,其基本设计思想是将GPS和SINS等导航设备输出的信息经过滤波器进行最优估计。但在采用Riccati方程更新协方差矩阵和计算Kalman增益过程中,截断误差随着迭代次数的增大而累积,破坏协方差矩阵的正定性和对称性,降低滤波器计算的数值稳定性,严重时导致组合系统故障发散。本文建立了Riccati方程一阶误差模型,从理论上分析截断误差对滤波器估计性能的影响,引入基于Bierman算法和Thorton算法的Kalman滤波器进行更新方法,解决了截断误差引起的滤波器数值稳定性的问题。通过强实时半物理仿真系统验证表明,相比于基于Kalman滤波器的系统,基于Bierman-Thorton算法的组合导航系统有更强的数值稳定性和较高的导航精度。
[Abstract]:As an important technique of positioning and attitude measurement, integrated navigation system is designed to estimate the information of GPS and sins through filter. However, in the process of updating the covariance matrix with Riccati equation and calculating the Kalman gain, the truncation error accumulates with the increase of iteration times, which destroys the positive definiteness and symmetry of the covariance matrix and reduces the numerical stability of the filter calculation. When serious, the failure of the combined system diverges. In this paper, the first-order error model of Riccati equation is established. The influence of truncation error on the estimation performance of the filter is analyzed theoretically. The Kalman filter updating method based on Bierman algorithm and Thorton algorithm is introduced. The problem of filter numerical stability caused by truncation error is solved. The verification of strong real-time semi-physical simulation system shows that the integrated navigation system based on Bierman-Thorton algorithm has better numerical stability and higher navigation accuracy than that based on Kalman filter.
【作者单位】：中国科学院长春光学精密机械与物理研究所;
【分类号】：TN96.2

【参考文献】