一种基于因子图的异步信息融合定位算法  

An Asynchronous Information Fusion Positioning Algorithm Based on Factor Graph

作　　者：张靖 王向东[1] 邓志宝[1] 张凌东 刘宇航[1] Zhang Jing;Wang Xiang-dong;Deng Zhi-bao;Zhang Ling-dong;Liu Yu-hang(Beijing Institute of Space Launch Technology, Beijing, 100076;PLA Rocket Force of Military Representative Office in the Capital Aerospace Machinery Company, Beijing, 100076)

机构地区：[1]北京航天发射技术研究所,北京100076 [2]火箭军驻211厂军事代表室,北京100076

出　　处：《导弹与航天运载技术》2019年第3期89-95,共7页Missiles and Space Vehicles

摘　　要：为了解决惯导/里程计/地磁匹配组合定位系统非等间隔数据融合带来拟合误差、信息源无法即插即用的问题,提出了一种基于因子图的异步多源组合定位算法。采用因子图关联了系统状态模型与观测模型,基于最大后验概率进行状态估计,充分利用了 GMNS 的非等间隔定位信息,消除了外推法的拟合误差。对组合定位算法进行了仿真,仿真结果表明:该算法的水平与高程定位误差分别减少到采用 GMNS 信息直接校正惯导位置方法的三分之一;进行了跑车试验,试验结果验证:该算法具备即插即用功能,提高了组合定位系统的可靠性。In order to solve the fitting error brought by the non-interval information fusion of the Inertial Navigation System(INS)/Odometer(OD)/Geomagnetic Matching Navigation System(GMNS) integrated positioning system and the information source cannot plug-and-play, an asynchronous multi-source information fusion positioning algorithm based on factor graph is proposed. The state model and observation model are linked by factor graph and the state estimation is based on the maximum posteriori probability. The non-interval positioning information of GMNS is fully utilized and the fitting error of the extrapolation method is eliminated. A simulation test is conducted whose results are presented that the level and elevation error of the algorithm are reduced to the third of the direct correction of the position of inertial navigation using GMNS information. An experiment is conducted whose results are presented that the algorithm has plug-and-play capability which improves the reliability of the integrated positioning system.

关 键 词：异步信息源 因子图 信息融合 即插即用 

分 类 号：V448.22[航空宇航科学与技术—飞行器设计]