似然迭代离差差分滤波算法的机动再入目标状态估计

State Estimation of Maneuvering Reentry Target Using Likelihood-Based Iterated Divided Difference Filter

针对使用离差差分滤波算法对机动再入目标状态估计时,不能充分用到最新量测信息,状态估计误差较低的情况,提出一种新的滤波算法——似然迭代离差差分滤波算法.该算法在二阶离差差分滤波算法的量测更新过程中采用Gauss-Newton方法不断逼近最大后验估计,且使用迭代状态估计值代替状态预测值,修正迭代公式,并使用确保产生的迭代序列向最大似然面移动的迭代终止条件.使用似然迭代离差差分滤波算法估计机动再入目标状态,蒙特卡罗仿真表明,该算法不仅提高了状态估计精度,而且还有很快的收敛速度.

The accuracy of the state esitmation is lower because the latest measurements are not fully used when the second-order divided difference filter is applied to estimate the state of maneuvering reentry target. A new algorithm, named likelihood-based iterated divided difference filter, is proposed. The Gauss- Newton method is used to approximate the maximun posteriori estimation in the measurement update process of the new algorithm, and state prediction is substituted with iteration state estimation. The iterate termination condition used makes the generated sequence move up to the likelihood surface. The proposed algorithm is applied to state estimation for maneuvering reentry target. Monte Carlo experiment results indicate that the new algorithm improves the accuracy of the state estimation and has a fast covergence rate.