Structural analysis and partitioning of dynamic process models for parallel state estimation

This paper presents a new decomposition algorithm that attempts to partition a large dynamic system into loosely coupled subsystems to be solved concurrently on network-based parallel computers (multicomputers) for the purpose of state estimation. This new technique is based on structural properties of the dynamic system and parallel computing considerations. In particular, the rate of convergence of a dynamic iterative solution scheme, employed for the coordination of subsystem integrations on different computer nodes, is used as a criterion for the selection of the best system partitioning among candidate partitions