多星载光学传感器系统误差极大似然配准算法

针对低轨星座协同探测弹道目标过程中存在系统误差的问题，提出多星载光学传感器系统误差极大似然配准（maximum likehood registration，MLR）算法。通过一阶Taylor近似对非线性量测转换线性化，推导出目标状态的误差协方差与卫星轨道定向、姿态角测量和传感器测量等随机误差的关系，并基于视线交叉获得观测在状态空间中的近似投影，从而将MLR算法扩展到低轨星座多光学传感器的误差配准。通过引入各类测量误差的先验信息对目标状态的误差协方差进行修正，利用期望极大化迭代，实现了对系统误差的无偏有效估计及目标轨迹的融合估计。仿真验证了所提算法的有效性，且配准性能优越。

Maximum likelihood registration for systemic error of multiple spaceborne optical sensors

Aiming at the problem of systemic error in detecting ballistic targets cooperatively via low earth orbit constellation, a maximum likelihood registration (MLR) algorithm is presented for systemic error of multiple spaceborne optical sensors. Firstly, the nonlinear measurement transformation is linearized by the first-order Taylor approximation, and the relations between the error covariance of target state and random errors of satellite orbit orientation, attitude angle measurement and sensor measurement are derived. Secondly, the approximate projection of measurement to the state space is obtained based on the line of sight crossing, thus the MLR algorithm is extended to the error registration for multiple optical sensors of low earth orbit constellation. Finally, by introducing the prior information of various measurement errors to correct the error covariance of target state, and using the iteration of expectation maximization, the unbiased effective estimation of systemic error and the fusion estimation of target trajectory are achieved. The numerical simulations demonstrate the effectiveness and the superior registration performance of the algorithm.