一种用于分散化滤波的融合算法及其在组合导航系统中的应用

A Fusion Algorithm of Decentralized Filter and its Application to Integrated Navigation System

提出了一种局部估计相关时的最优融合算法,其特点是加权矩阵是对角阵,即全局估计中的各分量分别是局部估计中相应分量的线性组合,在此融合算法中局部估计是最优的,并且不存在由主滤波器向子滤波器的信息反馈,因此具有较好的容错性。给出了最优加权矩阵的存在性和唯一性的证明,并通过与现有类似融合算法的对比,表明该算法具有计算量小的优点。最后,以组合导航系统为例作了仿真计算,结果表明该融合算法是可行的。

A fusion algorithm of correlated local estimates for integrated navigation system is given. The weighting matrices for fusing local estimates are diagonal, i.e., the global state estimate is only the linear combination of correspondent local state estimates. Here the local solutions are optimal and there is no information being fedback. So the fault tolerant is better. The existence and uniqueness of optimal diagonal matrices are proven. By comparing with the similar algorithm, it shows that these diagonal weighting matrices is very simple in calculation. Finally its application to an integrated navigation system is also discussed.