A Scalable Method for Semidefinite Programming Based Distribution System State Estimation

Abstract

Distribution system state estimation (DSSE) is crucial for real-time management of distribution networks. The weighted least squares (WLS) method is widely used for DSSE, via the Gauss-Newton algorithm. The Gauss-Newton algorithm often suffers from convergence issues when pseudo-measurements and virtual measurements are used. This motivates the development of semidefinte programming (SDP) based DSSE. This framework is improved by a convex iteration (CI) approach to obtain a high quality rank-one solution. But neither the standard SDP-DSSE nor the CI improvement is a scalable method. In this paper, a chordal decomposition based convex iteration (CDCI) approach using a quadratic cone (QC) is proposed. The proposed solution method is computationally scalable while obtaining a high quality rank-one solution. Simulation results on the IEEE 13-bus, IEEE 37-bus and IEEE 123-bus systems verify the performance of the CDCI approach.