论文部分内容阅读
研究了有数据丢包的带随机不确定参数的多传感器系统的分布式最优(线性最小方差)融合滤波问题。首先,引入虚拟噪声,将原系统转化为等价的参数确定的有丢包的新系统。然后,进行状态扩维,得到新系统的各子系统的扩维状态的滤波估计、滤波误差方差和滤波误差互协方差。根据扩维状态与原系统状态的关系,求出原系统状态的各局部滤波估计、滤波误差方差和滤波误差互协方差。利用线性最小方差意义下的矩阵加权最优融合算法,得到原系统的分布式矩阵加权最优融合滤波器。理论分析和仿真算例都表明,融合滤波器优于每一个局部滤波器。
The distributed optimal (linear minimum variance) fusion filtering problem in multi-sensor systems with data packet loss and random uncertain parameters is studied. First of all, the introduction of virtual noise, the original system into equivalent parameters determined by the new system with packet loss. Then, state expansion is performed to obtain the filter estimation, filter error variance and filter error covariance of each subsystems in the new system. According to the relationship between the extended state and the original system state, the local filter estimation, filter error variance and filter error covariance of the original system state are obtained. Using the matrix-weighted optimal fusion algorithm in the sense of linear minimum variance, the original system of distributed matrix-weighted optimal fusion filter is obtained. Both theoretical analysis and simulation examples show that the fusion filter is better than each local filter.