发射系下SINS/GPS/CNS组合导航系统联邦粒子滤波算法

传统的扩展卡尔曼滤波(Extended Kalman filter, EKF)算法应用于未来高超、空天飞行器的组合导航系统时，因其模型线性化展开会导致模型不准确，从而引起导航精度下降；采用蒙特卡洛方法来实现递推贝叶斯估计问题的粒子滤波(Particle filter,PF)算法能有效避免引入线性化误差，具有一定的优势。据此，针对高超、空天飞行器在发射过程中通常需要直接获得发射惯性系下的高精度导航参数的需求，提高发射惯性系下弹载组合导航系统滤波算法的精确性就尤为重要，PF滤波算法无需对非线性系统进行线性化展开即可直接实现对非线性系统的状态误差估计。为此，本文将PF滤波算法引入空天飞行器SINS/GPS/CNS多信息融合组合导航系统，设计了发射系下基于联邦滤波器的PF滤波算法，实现了对组合导航系统状态参数的直接建模估计。算法仿真结果表明，相较于发射系下SINS/GPS/CNS组合导航系统联邦EKF滤波算法，PF滤波算法有效提高了组合导航系统滤波精度。

SINS/GPS/CNS Integrated Navigation System Federal PF Algorithm in Launch Inertial Coordinate System

When the traditional extended Kalman filter (EKF) algorithm is used in integrated navigation system of future aircraft, it may lead to the inaccuracy of the model because of linearization and the decreasing of navigation precision. Particle filter (PF) solves Bayes estimation problem by using Monte Carlo method and can effectively avoid bringing in linearization error. Consequently, aimed at the requirement of high accuracy for the navigation system state parameters in launch inertial coordinate system, it is particularly important to improve the accuracy of filter algorithm for missile borne integrated navigation system. The PF algorithm can directly achieve an error estimation without linearization of nonlinear system. This paper brings PF algorithm into the SINS/GPS/CNS integrated navigation system.PF algorithm is designed based on federal filter and the navigation system state parameters are estimated directly. The algorithm simulation results indicate PF algorithm effectively improves filtering precision compared with EKF algorithm and is very suitable for integrated navigati on system.