序贯结构的Sage-Husa自适应滤波及其应用

在分析已有的Sage-Husa自适应滤波算法的基础上，本文首先推导了两种量测噪声自适应估计方法的等价性。为充分利用组合系统中已知的部分量测噪声参数，提高滤波稳定性和精度，研究了基于序贯结构的Sage-Husa自适应滤波算法；当组合系统测量噪声参数均为已知时，为降低算法复杂度，提高Sage-Husa自适应滤波的鲁棒性，加入协方差匹配的方法对序贯结构的Sage-Husa自适应滤波算法进行改进；通过在序贯结构下采用相应的信息融合策略，充分利用组合系统的输出信息。将两种算法分别应用于MIMU/GPS/磁强计组合系统中，基于跑车实验的离线数据分析表明，第一种滤波算法的滤波稳定性较标准自适应算法在滤波稳定性上有明显提高；第二种改进的滤波算法既降低了算法复杂度，又提高了抗野值效果，有效保持了组合系统在干扰状态下的导航精度。

The Sage-Husa Adaptive Filter of Sequential Structure Applied in Integrated System

Firstly, this paper deduces the equivalence of two kinds of noise matrix measuring methods. In order to fully utilize the partly known measurement noise parameters and enhance the stability of filtering, the Sage-Husa adaptive filter algorithm based on sequential structure is presented. When the noise characters of integrated system are known, to simplify the computation and improve the performance of Sage-Husa adaptive filter in rejecting outliers, covariance mapping method is introduced into the Sage-Husa adaptive filter algorithm based on sequential structure. The data fusion strategy is designed to fully utilize the output information of integrated system under the sequential structure. The two algorithms are applied into the MIMU/GPS/magnetometer integrated system. The off-line analysis based on vehicle data results show that stability of the first algorithm is obviously better than the standard structure adaptive filter; the second improved algorithm not only decrease the computation load, but also improve the performance of rejecting outliers and obviously keep the filtering accuracy.