论文部分内容阅读
X射线脉冲星导航由于需要大量的观测数据来确定天体方位,因此需要较长的滤波周期(500～1000s),系统的状态方程是非线性的。若采用扩展Kalman滤波(EKF),线性化时会产生较大的高阶截断误差,且Jacobian矩阵不易精确计算,导致定轨的精度较低。文中采用非线性滤波方法 -Unscented Kalman滤波(UKF),免去线性化过程,利用多个采样点的加权统计信息进行状态的预测。试验结果表明,UKF的滤波精度和稳定性明显优于EKF,但由于采样点均需进行轨道预测,计算所耗时间较长。当初始状态误差很大时,UKF误差减小稍慢于EKF,可在初始阶段时使用EKF滤波减小初始状态误差,之后使用UKF保证滤波的稳定性。
X-ray pulsar navigation requires a large number of observations to determine the orientation of the celestial body, which requires a longer filtering period (500 ~ 1000s) and the state equation of the system is nonlinear. If extended Kalman filter (EKF) is adopted, the linearization will produce larger high-order truncation errors, and the Jacobian matrix is not easy to calculate accurately, resulting in lower accuracy of orbit determination. In this paper, Unscented Kalman filter (UKF) is adopted to eliminate the linearization process, and the weighted statistical information of multiple sampling points is used to predict the state. The experimental results show that the filtering accuracy and stability of UKF are obviously better than that of EKF. However, the computation time is longer because of the orbit prediction required by the sampling points. When the error of the initial state is large, the error of UKF decreases slightly slower than EKF, and the initial state error can be reduced by using EKF filter in the initial stage. Afterwards, UKF is used to ensure the stability of the filter.