离散动态随机系统的信息融合滤波方法

针对离散随机动态系统的滤波问题,提出了基于信息融合估计的线性和非线性滤波方法。对于线性滤波,通过融合一步预测状态信息以及量测信息来获得系统状态的最优融合估计,并证明了它与标准卡尔曼滤波的一致性。基于非线性信息融合估计定理,推导出一种计算简单的迭代型非线性滤波方法。最后,还提出了信息融合滤波初值选取问题的解决方法。

Information fusion filtering for discrete dynamic stochastic system

The filtering methods based on information fusion estimation in linear or nonlinear systems was presented for the filtering problem in discrete dynamic stochastic system. For linear filtering, one step predictive information and state measurement information are used to obtain the optimal fusion estimate of the system state, and the uniformity between linear filtering and standard Kalman filtering was proved. Based on the nonlinear information fusion theorem, an iterative fusion estimation method with small computation cost was derived. Furthermore, a solution to the selection of initial filtering values was investigated.