一种改进的自适应USQUE组合导航姿态估计方法

An Improved Adaptive USQUE Integrated Navigation Attitude Estimation Algorithm

针对传统四元数无味卡尔曼滤波(unscented quaternion Kalman filter,USQUE)算法的量测噪声统计未知及时变引起滤波发散精度降低等问题,提出一种变分贝叶斯自适应四元数无味卡尔曼滤波算法(variational Bayesian-based adaptive USQUE,VB-AUSQUE)。通过变分贝叶斯高斯迭代近似估计,获取近似的量测噪声协方差矩阵滤波先验条件。仿真和舰载测试表明:在捷联式惯性导航系统/全球定位系统(strapdown inertial navigation system/global navigation satellite system,SINS/GNSS)组合导航系统中,VB-AUSQUE算法能有效减少系统量测噪声未知及时变问题对姿态估计精度的影响,相比常规算法具有更高的精度。

A variational Bayesian adaptive unscented quaternion Kalman filter(VB-AUSQUE)was proposed to solve the problem of reduced filter divergence accuracy caused by the unknown time variation of the measurement noise statistics of traditional quaternion unscented Kalman filter.The approximate measured noise covariance matrix filtering,a priori was obtained through the estimation of the variational Bayesian approximation.Simulation and shipboard tests show that in the combined strap-down inertial navigation system/global navigation satellite system(SINS/GNSS),the VB-AUSQUE system is the best solution for the combined navigation system.The AUSQUE algorithm can effectively reduce the influence of unknown time variation problem on the accuracy of attitude estimation,and the algorithm has a higher accuracy than the conventional algorithm.