Decentralized Low-Rank State Estimation for Power Distribution Systems

This article considers the low-observability state estimation problem in power distribution networks and develops a decentralized algorithm leveraging matrix completion methodology. Matrix has been shown to be an effective technique that exploits low dimensionality of system measurements recover missing information. can utilize approximate (linear) load flow model, or it used with no physical models network where information about topology line admittance is available. The direct application algorithms requires solving semi-definite programming (SDP) problem, which becomes computationally challenging for large networks. We therefore develop capitalizes on popular proximal alternating direction method multipliers (proximal ADMM). allows us distribute computation among different areas network, leading scalable algorithm. By doing all computations at individual control only communicating neighboring areas, eliminates need data sent central processing unit thus increases efficiency contributes goal autonomous illustrate advantages proposed numerically using standard IEEE test cases.