摘要: |
对因子图算法及其理论基础进行研究,使用消元算法分析了一元观测因子图模型的稀疏结构,得到了一种适用于该模型的增量推断算法;并对固定滞后平滑算法进行了研究,实现了滑窗在该增量推断算法中的应用;最后通过仿真验证了基于因子图的状态估计方法的估计能力、该增量推断算法在求解一元观测因子图模型的状态估计问题时运算速度的提升能力,以及滑窗维持该增量推断算法运算速度稳定的能力。 |
关键词: 非线性优化 因子图 变量消元算法 固定滞后平滑算法 |
DOI: |
|
基金项目:装备预研航天科工联合基金(6141B070503) |
|
Research on Real-time Operation of State Estimation Method Based on Factor Graph |
SUN Tao,ZHENG Xin,CHANG Qi,WU Liang-hua |
(Beijing Institute of Automatic Control Equipment, Beijing 100074, China;;China Aerospace Science and Industry Corporation Limited, Beijing 100048, China;;Equipment Project Management Center of the PLA Rocket Force Equipment Department, Beijing 100089, China) |
Abstract: |
The factor graph algorithm and its theoretical basis are studied, the elimination algorithm is used to analyze the sparse structure of the unary observation factor graph model, and the idea of incremental inference is combined to get an incremental inference algorithm suitable for the model. The fixed lag algorithm is researched, and the sliding window is applied in the incremental inference algorithm. Finally, simulation verifies the estimation ability of the state estimation method based on the factor graph, and it is also verified that the incremental inference algorithm is useful for increasing the calculation speed in solving the state estimation problem of the unary observation factor graph model, and the sliding window is able to maintain the stability of the calculation speed of the incremental inference algorithm. |
Key words: Nonlinear optimization Factor graph Variable elimination Fixed-lag smoothing algorithm |