A fully distributed optimal power flow algorithm for multi-regional DC systems based on ADMM

Abstract

To address the scalability limitations of traditional centralized optimal power flow algorithms， a fully distributed algorithm for solving optimal power flow in DC systems with large-scale distributed energy sources is proposed. Firstly， the non-convex optimal power flow model of DC systems is transformed into a second-order cone programming （SOCP） model using convex relaxation methods. Subsequently， a distributed algorithm based on the alternating direction method of multipliers （ADMM） is established to solve this model. The algorithm is then transformed by eliminating consensus variables into an improved fully distributed algorithm for optimal power flow in DC systems. Finally， in simulation tests on enhanced IEEE 13-node and 118-node systems， the algorithm proved capable of achieving a globally optimal solution for optimal power flow in radial and ring DC systems. Notably， this was accomplished without the necessity of a central coordinating unit to synchronize boundary consistency information across regions. Instead， the algorithm relies on parallel optimization within each region and the exchange of minimal boundary node information between adjacent regions.

Keywords

• optimal power flow
• distributed optimization
• convex relaxation
• socp
• admm