基于鲁棒增强的因子图多源信息融合算法

Multi-source Information Fusion Algorithm of Factor Graph Based on Robust Enhancement

为了解决组合导航系统中各传感器因受环境影响而产生观测值异常的问题,提出了一种基于鲁棒增强的因子图多源信息融合算法。首先,将组合导航解释为具有拓扑结构的因子图;其次,通过推理组合导航系统的因子图模型,对惯性导航系统(INS)、全球定位系统(GPS)和磁力计(Mag)进行信息融合,计算出导航状态的最优估计值;然后,在此基础上设计动态权重函数,合理动态地调整各因子的权重;最后,利用仿真试验将传统因子图算法和改进的因子图算法进行对比,结果表明:东、北轴向位置的误差分别为0.38、0.62 m,相比传统因子图算法,该算法位置精度提升近70%,拥有更好的导航性能和鲁棒性。

In order to solve the problem that the sensors in integrated navigation system produce abnormal observed values under the influence of environment,a factor graph multi-source information fusion algorithm based on robust enhancement is proposed.The integrated navigation is firstly interpreted as a factor graph with topological structure.Next,by reasoning the factor graph model of integrated navigation system,the information fusion of inertial navigation system(INS),global positioning system(GPS)and magnetometer(Mag)is carried out to calculate the optimal estimate of navigation state.Then,on this basis,the dynamic weight function is designed,and the weights of each factor are adjusted reasonably and dynamically.Finally,the traditional factor graph algorithm and the improved factor graph algorithm are compared by simulation experiments.The results show that the errors of the east and north axial positions of the improved algorithm are 0.38 m and 0.62 m respectively.And compared with the traditional factor graph algorithm,the position accuracy of this algorithm is improved by 70%,and it has better navigation performance and robustness.