一种航天器星光折射连续导航方法

针对星光折射航天器自主导航应用中观测缺失导航折射星造成误差上升甚至导航发散的情况，提出一种适用于航天器星光折射导航空白段的新方法。阐述了星光折射导航机理，给出了导航星观测窗口，进而设计基于神经网络的导航算法，该方法充分利用已有信息，有效预测并修正航天器状态信息，使星光空白段前后导航误差变化平稳，发挥星光折射间接敏感地平精度高的特点，保证了航天器高精度定位，且不需要添加硬件设备，算法简洁、实用。最后，通过计算机仿真校验了该导航方法的有效性。

A Continuous Navigation Method of Spacecraft Based on Stellar Refraction

When a satellite uses stellar refraction to conduct navigation, it often occurs that the navigation refraction stars can’t be observed because of the influence of the celestial bodies such as the Earth, Sun and Moon. Thus the accuracy of locating will decrease a lot. Moveover, the filtering of navigation can be diverging. This paper proposes a new method based on neural network compensating navigation data in the blank section of starlight. The method utilizes the related properties between the correlation of the navigation errors and status errors. It establishes three layers of the neural network system, which could effectively predict and correct the information of the satellite status and could ensure that the navigating errors change smoothly around the starlight blank section. Then the high-precision navigation for satellite can be achieved. This paper constructs the influence models of the celestial bodies and describes the mechanism about how the blank section of starlight occurs and the scheme of designing the neural network system. Finally, the efficiency and high accuracy of compensating the blank section method is verified via simulation.