运用信息融合式高阶UKF的微小卫星姿态确定算法

Attitude Determination Algorithm for Micro-satellite Based on High-order UKF Using Information Fusion

为提高微小卫星微型低成本姿态敏感器的姿态确定精度,基于磁强计/太阳敏感器/陀螺仪的姿态敏感器配置以及无迹卡尔曼滤波方法(Unscented Kalman Filter,UKF),设计了一种基于高阶UKF算法并且融合磁强计与太阳敏感器观测信息的微小卫星姿态确定算法.为提高系统状态方程非线性函数的一步预测精度,采用基于五阶UT变换的高阶UKF算法,增加了Sigma采样点数量,提高了系统状态预测精度.单一观测向量滤波算法不能同时满足多个不同量纲观测数据,本文提出一种同时利用两个观测向量的信息融合式滤波算法,根据磁强计和太阳敏感器的观测信息,通过卡尔曼滤波原理中的增益计算,分别得出地磁矢量和太阳矢量对应的卡尔曼增益信息.采用高斯概率密度准则进行信息融合,进而完成预测值的修正,得到同时满足磁强计以及太阳敏感器观测需求的四元数估计值,降低了观测误差的影响.仿真分析验证了算法的优越性.

To improve the attitude determination accuracy of miniature low-cost attitude sensors,a micro-satellite attitude determination algorithm that combines magnetometer and solar sensor observations was designed,based on high-order Unscented Kalman Filter(UKF)and the attitude sensor configuration of magnetometer/solar sensor/gyroscope.Firstly,in order to improve the one-step prediction accuracy of the nonlinear system state equation,the high-order UKF algorithm with fifth-order UT transformation was used to increase the number of Sigma sampling points and improve the system state prediction accuracy.Secondly,owing to the shortcoming of the single observation vector filtering algorithm that couldn't coordinate multiple observation data with different dimension simultaneously,an information fusion filtering algorithm using two observation vectors was proposed,which was based on the observations of the magnetometer and the solar sensor.The Kalman gain was obtained by the information corresponding to the geomagnetic vector and the solar vector through the gain calculation of the UKF algorithm.Consequently,the Gaussian probability density criterion was used to fuse the Kalman gain information,and the fused information was used to correct the value of the one-step prediction.Therefore,this algorithm reduced the observation errors of the attitude quaternions.Finally,the simulation results proved the availability of the proposed method.