A Decentralized State Estimation Algorithm for Building Electrical Distribution Network Based on ADMM

Abstract

Large-scale building electrical distribution network calls for precise data. To ensure the stable operation and lower energy consumption of the distribution network, these operation data need to be checked frequently by State Estimation (SE) algorithm. Traditional state estimation methods are inefficient because the problems are non-convex and Kirchhoff's laws are global. In this paper, a decentralized SE method is proposed for an intelligent building architecture named Insect Intelligent Building (I <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> B) system. This method accomplishes the state computation through the cooperation between neighbor intelligent nodes, just like the insects in the nature world. It decomposes the traditional centralized state estimation problem into many sub-problems, which could be solved in the intelligent nodes with the constraint based on the circuit principle. In this paper, we provide a parallel iterative algorithm for these sub-problems by using Alternating Direction Method of Multipliers (ADMM). Unlike traditional state estimation algorithms which require global data, our decentralized algorithm only requires the information from neighbors. Therefore, it helps to lower the computation cost and facilitates the iterative process of state estimation. We conduct experiments on the building electrical distribution network using proposed methods. Simulation results demonstrate the scalability and optimality of the proposed algorithm.