基于对偶四元数的惯性/卫星/天文组合导航系统改进联邦滤波方法

传统惯性/卫星/天文组合导航系统联邦滤波中主滤波器和子滤波器状态量保持一致，而天文姿态信息仅对部分状态具有可观测性。常用的惯性/天文组合模式需将高精度的天文姿态信息进行坐标转换，从而降低组合导航精度。针对上述问题，提出了一种基于对偶四元数的惯性/卫星/ 天文组合导航系统改进联邦滤波方法，通过可观测性分析对子滤波器进行降维，能在保证导航精度的同时提高系统实时性。基于对偶四元数的降维惯性/天文组合方法可直接利用天文姿态信息进行组合，实现了天文高精度信息的有效利用，且避免将不可观测状态进行反馈修正从而提高组合精度。仿真结果表明，提出的改进联邦滤波方法能获得与集中式滤波相当的精度，在卫星信号丢失或故障情况下，相比于传统惯性/天文组合具有更优的性能。

An Improved SINS/GNSS/CNS Federal Filter Based on Dual Quaternions

In traditional SINS/GNSS/CNS federal filter,the master filter and the local filters have the same state variables.However,the attitudes information output by CNS is observable for some state components in state vector.In common SINS/CNS integrated mode,the output information of CNS needs to be converted to navigation coordinate frame,which may decrease the accuracy of integrated navigation system.For the problems above,an improved SINS/GNSS/CNS federal filter based on dual quaternions is proposed.The dimension of local filter is decreased by the observability analysis of state vector,which assures the navigation accuracy and improves the real-time performance at the same time.The improved SINS/CNS integrated mode based on dual quaternions can utilize the attitudes information of CNS to integrate directly and avoid using those unobservable state components to correct SINS.Thus,the effective use of high precision CNS information is realized.The simulated results show that the proposed method has the same navigation accuracy as centralized Kalman filter,and can obtain better navigation performance when GNSS signal is lost.