提出一种适用于低轨道卫星的自主导航方法。与以往采用星光角距作为观测量的方法不同，该方法利用卫星平台中常见的星敏感器和地磁敏感器确定卫星实时轨道参数；通过对星敏感器的观测数据进行适当转换，将卫星位置单位矢量观测方程转换为线性化方程；对地磁场强度与轨道高度的关系进行拟合，利用磁强计观测数据求取地心距；采用拉格朗日差值算法确定卫星初轨，为滤波器提供初值。由于二体轨道动力学模型的一阶线性化近似引入的系统模型误差影响了滤波器性能，而一阶导数函数值的线性组合可替代高阶导数的函数值，并降低计算量，所以本文在扩展KALMAN滤波器的设计过程中，对状态方程进行了高阶线性化处理，从而提高了滤波器状态方程的离散化精度。最后，用仿真实验验证了这种低轨卫星轨道确定方法的有效性与实用性。

An autonomous navigation method suitable for low orbit satellites is proposed. Different from traditional method that takes a start light angle as the observed variable, the method determines the real time orbital parameters by a start sensor and a geomagnetism sensor, and converses the unit vector observation equation of satellite position into linear equations by converting the observation data from the start sensor appropriately. With fitting the relationship of magnetic field strength and orbit height, it obtains the geocentric distance by using a magnetometer. Moreover, it determines the satellite initial orbit by Lagrange difference algorithm to provide initial value for the filter. As the performance of the filter is effected by the system model error caused from the firstorder linear approximation of two body dynamic orbit model, and the high order derivative function values can be replaced by a linear combination of one order derivative function values, this paper processes the state equation by a higher order linearity in design of the KALMAN filter to reduce the calculation. By proposed method, the discretized accuracy of state equation for the filter is improved greatly. Finally, a simulation experiment is performed, which verifies that the method is effective and feasible.