基于变分贝叶斯的自适应鲁棒滤波算法

Adaptive Robust Filtering Algorithm Based on Variational Bayesian

针对组合导航姿态估计中,观测同时受到野值与时变观测噪声影响的问题,构造一种基于变分贝叶斯的自适应鲁棒滤波算法。该算法可以有效地解决自适应与鲁棒滤波策略的矛盾,利用变分贝叶斯近似估计变换的观测噪声,在变分贝叶斯的滤波框架内,利用Huber滤波鲁棒化方法处理连续野值。在组合导航姿态估计试验中,验证了该算法具有良好的自适应与鲁棒性,并能够保持较高的估计精度。

In this paper,an adaptive robust filtering algorithm based on variational Bayesian method is proposed to solve the problem of the simultaneous observation of outliers and time-varying noises in the attitude estimation of integrated navigation.The algorithm can effectively solve the contradiction between the adaptive and robust filtering strategy,using variational Bayesian approximation to estimate the observation noise transformation,and deal with continuous outliers by using Huber filter robust method in the variational Bayesian filtering framework.In the integrated navigation attitude estimation experiment,it is proved that the algorithm has good adaptability and robustness,and maintains high estimation accuracy.