Distributed Parallel Filtering for Spatially Interconnected Systems

A distributed parallel filtering algorithm is proposed for spatially interconnected systems.The characteristic of this algorithm is that multiple computing units are used to estimate the states of all the subsystems.Only utilizing the output of current subsystem and the estimation results from the adjacent computing units,each computing unit estimates the states of current subsystem and sends the results to its adjacent computing units.In on-line recursive computation,the computational complexity of this algorithm is just equivalent to that of Kalman filter for a single subsystem.Simulation results show that the filtering precision of this algorithm is almost the same as that of the optimal lumped Kalman filter.The steady-state estimation error between these above two algorithms is of only about 5% difference.