Fully Decentralized Optimal Power Flow Of Multi-Area Interconnected Power Systems Based On Distributed Interior Point Method

Abstract

The Optimal Power Flow (OPF) model represents the problem of determining the bestoperating levels for electric
power plants in order to meet demands given throughout a transmission network, usually with the objective of
minimizing operating cost. Traditionally, the optimal power flow (OPF) problem is solved in a centralized manner.
However, with continuous expansion of the scale of multi-area interconnected power systems, realisticapplications of
the centralized method face additional challenges. To date, a number of decomposition techniques have been
proposed to tackle decentralized OPF problems such as Lagrangian relaxation, Auxiliary Problem Principle (APP), the
Alternating Direction Multiplier Method (ADMM), Benders decomposition, dual problem formulation etc., In this
paper, we propose a fully decentralized OPF algorithm for multi-area interconnected power systems based on the
distributed interior point method, where solving the regional correction equation was converted into solving a
parametric quadratic programming (QP) problemduring each Newton-Raphson iteration. Our proposed method
utilizes a unidirectional ring communication graph for information sharing among areas, to eliminate the upward and
downward communication among cliques. In addition, our method not only solves loosely coupled problems, but can
also solve other complex distributed computing problems ofpower systems (e.g. non-convex optimization problem)
in a fully decentralized manner. The proposed approach is fully decentralized without the need of a central authority
to compute the cliques or update the multipliers, and no parameter tuning is required. It is robust to network
partitioning. Namely, its convergence remains stable regardless of how the system is partitioned. Furthermore, it is
applicable to any networks, including real multi-area power systems. This is a novel approach to tackling decentralized
OPF problems. Furthermore, this decentralized algorithm enjoys the same convergence performance and accuracy as
the centralized interior point method.