基于QR分解的自适应差分滤波在SINS大方位失准角初始对准中的应用

A QR Factorization-based Adaptive Divided Difference Filter for Initial Alignment of Large Azimuth Misalignment of SINS

针对差分滤波算法存在数值稳定性差及系统噪声统计特性不准确导致滤波性能下降的问题,提出一种改进的差分滤波算法。新算法将矩阵QR分解及Cholesky分解因数更新技术引入,并结合一种新的噪声估计方法及选择更新策略,有效地改善了系统噪声统计特性不准确带来的滤波性能下降问题,增强了传统差分滤波算法的数值稳定性。通过在捷联惯导系统大方位失准角初始对准中的应用,仿真结果表明了新算法的有效性和鲁棒性。

The divided difference filter for nonlinear state estimation suffers from numerical instability and the problem of degraded performance of the filter due to the incorrect statistics of the system noise, an adaptive difference filter is developed. The new algorithm, combining with the matrix QR decomposition, Cholesky decomposition factor updating, noise estimator and selective update strategy, efficiently im-proved the above problem. Test on the strapdown inertial navigation system （SINS） of initial alignment of large azimuth misalignment and simulation results show the efficiency and robustness of the proposed method.