Diffusion widely linear adaptive estimation of system frequency in distributed power grids

Abstract

Motivated by the need for decentralized estimation in the electricity grid, we apply the diffusion augmented complex least mean square (D-ACLMS) algorithm to estimate the frequency of a three-phase power system under balanced and unbalanced voltage conditions in a distributed setting. In addition to using local measurements, the diffusion adaptation strategy allows nodes to share their frequency estimates with neighbouring nodes which enables the network to exploit multiple measurements of the voltage signal. The analysis provides convergence conditions of the D-ACLMS and also gives the bounds on parameter values which guarantee that the diffusion type of adaptation stabilizes the network. Simulations under both balanced and unbalanced conditions support the analysis.