组合导航系统中一种基于IMM-Kalman的数据融合方法  

Data fusion method based on IMM-Kalman for an integrated navigation system

作　　者：田易[1,2] 阎跃鹏 钟燕清[1] 李继秀 孟真 TIAN Yi;YAN Yuepeng;ZHONG Yanqing;LI Jixiu;MENG Zhen(Institute of Microelectronics of the Chinese Academy of Sciences,Beijing 100029,China;University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院微电子研究所,北京100029 [2]中国科学院大学,北京100049

出　　处：《哈尔滨工程大学学报》2022年第7期973-978,共6页Journal of Harbin Engineering University

基　　金：中国科学院战略性先导科技专项项目(XDA22020102)。

摘　　要：针对INS/GNSS组合导航系统中对传感器随机零偏动态估计的需求,本文提出了一种基于IMM-Kalman的数据融合方法。该算法基于IMM框架,通过模型概率更新,进行自适应系统模型选择,实现对传感器零偏的动态最优估计,提高组合导航系统数据融合精度。仿真验证表明:与常规算法相比,改进算法可以实现对传感器随机零偏的动态估计,提高姿态解算精度,将俯仰角和滚转角误差由0.26°和0.23°分别降至0.11°和0.04°;同时仿真了卫星失锁的情况,改进算法的应用,使INS独立导航时位置累积误差仅为21.8 m,较常规方法的42.2 m减少了近50%,进一步证明了算法的有效性。This study proposes a data fusion method based on an interactive multimodel Kalman filter(IMM-Kalman)to meet the demand for the random zero-offset dynamic estimation of sensors in the INS/GNSS integrated navigation system.The algorithm adaptively selects the system model based on the IMM framework by updating the model probability,realizing the optimal dynamic estimation of sensor zero bias,and improving the data fusion accuracy of the integrated navigation system.Simulation results showed that the improved algorithm could dynamically estimate the sensor′s random zero bias and improve the attitude solution accuracy compared with the conventional algorithm.The pitch angle and roll angle errors were reduced from 0.26°and 0.23°to 0.11°and 0.04°,respectively.Simultaneously,the situation of a satellite losing its lock was simulated.The total position error was only 21.8 m.When INS was navigated independently using the improved algorithm,the distance traveled was nearly half that of the conventional method of 42.2 m.This further demonstrates the improved algorithm′s increased effectiveness.

关 键 词：组合导航 惯性导航系统 微机电系统 陀螺仪 加速度计 交互多模型 卡尔曼滤波 随机零偏 数据融合 

分 类 号：V249.32[航空宇航科学与技术—飞行器设计]