基于粒子滤波的微小卫星姿态确定算法

为了提高非线性卫星姿态控制系统的滤波性能,在建立了采用磁强计及太阳敏感器的卫星姿态模型的基础上尝试了新兴的粒子滤波(PF)算法对卫星系统进行姿态估计,进而对采用矢量观测的三轴稳定卫星的姿态确定问题进行了滤波算法的实时仿真,并将四元数转换成旋转矢量引入了粒子滤波算法,最后给出了卫星模型在不同粒子数目下的滤波性能比较,并在系统初始误差较大的情况下将粒子滤波算法与EKF滤波算法进行了滤波性能的对照。仿真结果表明,粒子滤波算法对粒子数目具有明显的依赖性,但是当粒子达到一定的数目时,粒子滤波的精度以及滤波稳定性都可以得到保证,尤其是在系统初始误差较大的情况下粒子滤波算法更显示了其优于EKF算法的滤波性能。

Particle filtering-based algorithm for micro-satellite attitude determination

In order to get more effective filter performance for non-linear system, a satellite model using magnetometers and sun sensors was established, and a Particle Filter Algorithm was used to estimate the attitude. Concerning the problem of three-axis stabilized satellite attitude determination from vector observation, the author treated quaternion as rotation vector in the Algorithms, then the filter performance comparison in two different mumble of particles was given. The simulation results approve that the capability of the algorithm relies on the number of the particles, however, the precision and stabilization can also get the required target with enough particles. At last, the paper compares the Particle Filter with EKF algorithm when the initial error is large. The estimate result validates the advantage of Particle Filter.