含状态等式约束的分布式估计融合

在假设各传感器间观测噪声不相关的基础上,将最优线性无偏估计意义下的分布式最优估计融合公式推广至含状态等式约束的分布式估计融合问题中,并在此约束条件下,先将该约束系统直接转换为投影系统,再利用投影系统中的Kalman滤波估计,分别讨论了两种信息融合方式:中心式与分布式融合。进一步地指出中心式估计融合优于分布式估计融合,并在分布式融合结构下,得到了约束的Kalman滤波估计优于投影的Kalman滤波估计的结论。通过数值模拟论证了两种融合方式下的性能差异。

Distributed Estimation Fusion for Systems with State Equality Constraints

Under the assumption of independent observation noises across sensors,an optimal estimation fusion formula for a general distributed system in the sense of the best linear unbiased estimation（BULE） is extended to the distributed estimation fusion for systems with state equality constraints.These systems can be directly transformed to the projected systems,by using the constrained Kalman filtering for the projected system,such two estimation fusion architectures as centralized and distributed are discussed.Moreover,the conclusions that the centralized estimation fusion outperforms distributed,and the constrained Kalman filter estimation fusion is better than projected Kalman filter estimation fusion are acquired.Finally,a numerical simulation demonstrates the performance difference between these two cases.