改进的强跟踪平方根UKF在卫星导航中应用

针对应用于受不确定性干扰和噪声影响的卫星自主导航系统中的无迹卡尔曼滤波（unscented Kalman filter,UKF）存在估计精度低、跟踪性能差和鲁棒性弱等缺陷,提出一种改进的强跟踪平方根UKF(strong tracking square-root UKF, STSRUKF)导航方法。该方法中利用星敏感器和光学导航相机设计出导航方案,并通过转换方程将间接量测量转换为观测量。针对平方根UKF（square-root UKF, SRUKF）在高阶系统中因为sigma点的零权值系数是负的或者数值计算误差太大时而可能造成滤波器发散问题,采用一种改良的平方根分解方法,改善了滤波器的稳定性。同时,基于强跟踪滤波器理论（strong tracking filters, STF）,引入多重自适应衰减因子调节协方差矩阵,使得滤波器具有强跟踪能力和克服系统模型不确定的鲁棒性,改善了滤波器的估计精度。将该方法应用于卫星自主导航系统中,实验仿真结果表明,相对于平方根UKF和STF,该方法不仅保证了系统的可靠性,还提高系统的导航精度和改善系统的鲁棒性及跟踪能力。

Satellite autonomous navigation filtering algorithm based on improved strong tracKing square-root UKF

For the satellite autonomous navigation system subjects to model uncertainties, external disturbances and noises, the unscented Kalman filter (UKF) method has low accuracy, poor tracking ability and poor robustness. An improved strong tracking square root unscented Kalman filter (STSRUKF) based autonomous navigation method is proposed. For the navigation purpose, star sensors and optical navigation cameras are used in this method, and the indirect measurement vector is transformed to observables through a transition equation. To avoid the problem that negative zero weights of sigma points and great calculation errors in square-root UKF (SRUKF) design for high order systems, a modified square-root decomposition method is applied for the SRUKF design to improve the stability of the SRUKF. In addition, based on strong tracking filters (STF), multiple adaptive fading factors in adjustment covariance matrix are adopted so that the STSRUKF has better tracking ability, better robustness against model uncertainties and better estimation accuracy. Finally, the STSRUKF based method is applied to the satellite autonomous navigation systems, and simulation results are provided to verify the effectiveness and practicability of the proposed approach.