A Real-time Alternating Direction Method of Multipliers Algorithm for Non-convex Optimal Power Flow Problem

The high penetration rate of smart devices like storages bring new challenge and complexity to the optimal power flow (OPF) problem. This problem is generally nonconvex and difficult to calculate in the real-time scenarios. This paper will introduce a fully distributed approach by combining the alternating direction method of multipliers and proximal alternating minimization method. This approach contains two parts, one is a basic distributed algorithm for the offline scheduling with constant network data. The another extended one is by using the short-term load forecast as its initial input, then when the real-time data is given, the solution of OPF problem can get converged faster than the basic algorithm. Both algorithms aim to provide a high feasible solution for the realtime grid operation. These algorithms are simulated on test network with batteries to analyze their performance and the effect of the changes in the parameters in these algorithms.