确定编队卫星相对轨道的容错UKF方法

Autonomous Determination of Relative Orbit for Satellites Formation Flying Using Fault Tolerant UKF

针对相对轨道测量设备出现连续野值引起的滤波误差偏大及发散问题,提出了基于残差正交性检测的容错UKF算法。介绍了非线性滤波在相对轨道自主确定中发展历程和容错UKF的原理、滤波算法和计算流程。根据测量设备出现连续成片野值时残差序列服从以野值为均值的正态分布这一特性,提出了基于残差正交性进行野值检测的思路,通过加权一个以残差方差的迹为参数的活化函数,构造在线修正的UKF容错滤波器,抑制野值对新息序列正交性的影响。在STK软件设定两颗卫星的轨道根数,生成相对运动的位置和速度,通过坐标变换,加入正态分布的误差和连续野值,构造卫星间距离、仰角和方位角测量值。根据C-W相对运动方程和空间几何建立状态方程和测量方程,编制基于残差正交性检测的容错UKF滤波器的相对轨道确定程序。通过仿真计算,评估了连续野值对常规UKF和容错UKF轨道确定算法的影响,证实了容错UKF算法可以完成对连续野值的检测及在线修正功能,有较强的鲁棒性和稳定性,可以应用于工程实践。

When the measured values influenced by continuous outliers in the relative orbit measure, filtering error becomes intolerable, even diffuses. Based on residual orthogonal UKF（ Unscented Kalman Filer） fault-tolerant filtering method is presented. The development of nonlinear filtering in autonomous determination of satellite＇ s relative orbit, the fault-tolerant UKF method＇ s theory, filtering algorithm and processes are introduced. When the measured values influenced by continuous outliers, the residual error sequence＇ s is normal distribution whose mean value is outliers. An idea to detect outliers based on residual orthogonal is presented, introducing an activate function whose argument is variance＇ s trace, then making an online-correction UKF fault-tolerant filtering, which can reject the outliers＇ effect to new information sequence＇ s orthogonality. Set two satellite＇ s orbit element in the STK（ Satellite Tool Kit） , then generate the value of relative position and relative velocity, after coordinate transformation, add in the normal distribution error and continuous outliers, then get the measured value of the distance, elevation and azimuth between two satellites. A nonlinear system including state and measure equations is established, basing on C-W equations and space geometry. Relative position and velocity between two satellites is estimated according residual orthogonal check and fault-tolerant UKF method. The simulation evaluates the influence of outliers in states estimation using fault-tolerant and common UKF. Fault-tolerant UKF can online check out outliers and modify the estimate state according the information of them. So the Fault-tolerant UKF method is robust and stabilized.