Abstract: The performance of the Kalman filter in SINS/GPS integrated navigation deteriorates drastically due to the fact that GPS signals are sheltered as they lead to the unexpected error of the measurement. Aiming at this problem, a modified adaptive Kalman filter is proposed on the basis of analyzing the adaptive Kalman filter based on the online estimation of the covariance of measured noise. Based on adjusting the covariance matrix of state predication error by faulty innovation, the new method solves the problem of serious non-conformance between the covariance of innovation and the actual. The result of simulation experiment proves the effectiveness of the novel method compared with the Kalman filter, the adaptive robust Kalman filter and the adaptive Kalman filter.

Key words: integrated navigation, adaptive Kalman filter, innovation, covariance

摘要： 全球定位系统（GPS）因信号受到遮挡和干扰而产生观测量突然失准，使捷联式惯性导航系统（SINS）/GPS组合导航的卡尔曼滤波器性能急剧下降。针对上述问题，提出了一种改进的自适应卡尔曼滤波的方法，通过失准时的新息对先验状态均方误差阵进行自适应调节，解决了新息协方差与实际严重不符的问题。仿真实验中，对比了传统的卡尔曼滤波、自适应卡尔曼滤波、自适应抗差卡尔曼滤波和改进的自适应卡尔曼滤波的估计性能，证明了所提出算法的有效性。

关键词: 组合导航, 自适应卡尔曼滤波, 新息, 协方差