航天器转动惯量参数在轨辨识的最优激励

根据矩阵分析理论研究了转动惯量参数的可辨识性问题,从姿态动力学方程导出关于未知参数的线性回归模型,将回归矩阵的条件数作为参数可辨识度的定量指标,以可辨识度最大为目标函数建立最优控制模型;应用解最优控制问题的伪谱法,计算控制力矩陀螺的指令轨线,使得惯量参数的辨识精度和收敛速度显著提高;在仿真中采用双无迹卡尔曼滤波算法进行参数辨识,结果验证了所提出的最优激励设计方法可以改善参数辨识性能,同时该方法对先验信息误差具有鲁棒性。

Optimal excitation trajectories for spacecraft’s moment of inertia parameters on-orbit estimation

The identifiability of moment of inertia parameters is studied base on matrix analytical theory. Firstly, a regression model that is linear in term of the unknown parameters is derived from dynamic equations; the quantifiable index of identifiability degree is defined as the reciprocal of condition number of regression matrix. Secondly, an optimal control problem is built, identifiability degree index as the cost function. Thirdly, the pseudospectral optimal control is employed to solve the resulting problem and generate optimal control moment gyro exciting trajectories for parameter identification. Finally, the Dual-unscented Kalman filter is used to identify the unknown parameters. Simulation results demonstrate that the optimal excitation shorten the convergence time and give more accurate estimation. the robustness of the proposed approach for synthesize optimal excitation is verified for the case where a inaccurate priori knowledge of unknown parameters.