随机奇异系统分布式最优分量融合滤波器

Distributed optimal component fusion filter for stochastic singular systems

应用射影理论,基于奇异系统典范型分解,对带相关噪声的单传感器随机奇异系统,给出一种新的递推滤波器;当系统带有多个传感器时,基于线性最小方差标量加权的分量融合算法,给出了多传感器分布式最优分量融合滤波器.融合估计的每个分量分别由局部估计的相应分量按标量加权融合获得,它只需并行计算一系列标量权重.可改善各局部估计的精度和减小计算负担.推得了随机奇异系统任两个局部估计之间的滤波误差互协方差阵.仿真例子验证了其有效性.

Applying projection theory and a decomposition in canonical form for singular systems, a new recursive filter is given for stochastic singular systems with correlated noises measured by single sensor. A multi -sensor distributed optimal component fusion filter for stochastic singular systems with multiple sensors is proposed based on the component fusion algorithm weighted by scalars in the linear minimum variance sense. Each component of the fusion estimator is obtained by scalar weighting fusion from the corresponding components of local estimators, respectively. It only requires in parallel a series of computations of the scalar weights, and can improve accuracy of local estimator and reduces the computational cost. Furthermore, the filtering error cross -covariance matrix is derived between any two sensor subsystems of stochastic singular systems. A simulation example verifies its effectivess.