卫星自主导航中的偏差分离鲁棒滤波

A Separate Bias Robust Filter in Autonomous Navigation for Satellite

研究卫星自主导航中的滤波算法。在以地心矢量为观测量的卫星自主导航系统中,由于观测模型不准、测量设备精度有限等因素,使得系统的观测量存在多种偏差,最终问题可抽象为观测量中含有偏差的离散线性系统的状态估计问题。把观测量中难以建模的复杂时变偏差看作不确定项,用一个模型误差界参数来描述不确定性的影响,由此设计鲁棒卡尔曼滤波器。进一步,把常值偏差与状态进行分离,分别构造无常偏的鲁棒滤波器和针对常偏的鲁棒滤波器,最终得到偏差分离鲁棒滤波算法。仿真表明该算法可以有效地抑制系统中的不确定性偏差,明显提高了卫星自主导航的精度。

This paper studies the filter in autonomous navigation for satellite. Due to the inaccuracy of sensor and system model, there always exist various biases in the autonomous navigation system with geocentric vector as its measured variable, and the question can be looked as state estimated question for discrete linear system with disturbed measurement. This paper takes the complicated perturb as uncertainty, uses a parameter to characterize the effect of uncertainty, and designs a robust Kalman filter. Moreover, this paper separates the const bias from states, and designs robust filter for states and bias respectively. Finally we get a separate bias robust filter. The simulation shows its validity, and the precision of autonomous navigation for satellite is improved.