Distributed fusion filtering algorithm for large-scale power systems with cross-correlated noises

This paper studies the distributed problem for dynamic state estimation of large-scale power systems with cross-correlated noises. The considered system is assumed to be topologically observable, and can be divided into several non-overlapping subsystems, which interact with each other to form a multi-area interconnected power system. Based on the projection theory, we propose a distributed fusion filtering algorithm, in which the edge measurements can be transformed into the local measurement by using the neighbouring information, and each subsystem gets the local state estimate by an iterative method at each time instant. The proposed algorithm converges in finite steps when the considered power system is acyclic, and the estimation accuracy is similar to that of the centralised fusion filter, which is optimal. The main contribution of this paper is that our proposed algorithm improves a known distributed state estimation algorithm to reduce the computational cost and decrease the storage space. Simulations are tested on the IEEE 118-bus system under various operating conditions to evaluate the performance of the distributed fusion filter, which verify the effectiveness and feasibility of our proposed algorithm.