两种网络约束下NNSs分布式有限时域FIR融合估计器

Decentralized receding horizon FIR fusion estimator for state estimation of NNSs with two network constraints

针对带宽受限和网络拓扑随机切换约束下组网导航系统(NNSs)的分布式位姿状态估计问题,建立网络约束综合作用下的离散组网滤波增广系统模型,提出一种分布式有限时域FIR融合估计算法.目标节点从邻节点集合中接收经量化效应和饱和效应处理后的普通数据包和跟踪数据包,在给出无偏约束条件的前提下,以使得状态估计值的估计误差满足最小方差为准则,充分考虑有色噪声的影响,设计有限时域FIR估计器及其差分形式,通过普通数据包得到目标节点状态的区域估计值,建立系统本地状态估计的统一机制.同时,考虑网络约束,将跟踪数据包引入系统的融合过程,在以均方准则确定时变加权矩阵的前提下,给出最优权值所满足的线性代数方程以及融合误差协方差的差分形式,将目标节点状态的区域估计值与各邻节点随机发送的协作估计值加权融合,得到目标节点状态的全局融合估计值.最后通过算例仿真验证算法的有效性.

In order to solve the problem of distributed pose state estimation for networked navigation systems(NNSs) with bandwidth constraints and network topology random switching constraints,a distributed finite time-domain FIR fusion estimation algorithm is proposed based on the model of the the discrete-time filtering augmented system under the above network constraints.The target node receives the general data packet and tracking data packet processed by quantization effect and saturation effect from the set of adjacent nodes.Under the premise of unbiased constraint,the estimation error of state estimation meets the minimum variance as the criterion,and the influence of colored noise is fully considered.The finite time domain FIR estimator and its difference form are designed,the target is obtained through the ordinary data packet,and a unified mechanism of local state estimation is established.At the same time,considering the network constraints,the tracking data packet is introduced into the system fusion process.Under the premise of determining the time-varying weighting matrix based on the mean square criterion,the linear algebraic equation satisfied by the optimal weight and the difference form of fusion error covariance are given.The regional estimation value of the target node state is weighted fused with the cooperative estimation value sent by each neighbor node randomly to obtain the global fusion estimator of the target node state.Finally,an example is given to verify the effectiveness of the algorithm.