Online Distributed Algorithm for Optimal Power Flow problem with Regret Analysis

TL;DR

This paper introduces a distributed online algorithm for the DC-Optimal Power Flow problem that adapts to uncertainty in renewable supply and market prices, using regret analysis to evaluate performance.

Contribution

It proposes a novel distributed online primal-dual algorithm for DC-OPF under uncertainty, with theoretical regret bounds and practical demonstration on IEEE-14 bus system.

Findings

Achieves sub-linear regret bounds under certain conditions.

Effectively manages constraint violations in a distributed setting.

Demonstrates good performance on IEEE-14 bus system.

Abstract

We investigate the distributed DC-Optimal Power Flow (DC-OPF) problem for a dynamic and uncertain environment. The unpredictable supply of renewable resources and varying prices of the electricity market are a few factors responsible for the uncertainty. We propose to address this problem using the framework of online convex optimization, where the cost functions are not known apriori because of the uncertainty and are revealed only incrementally over time. We also consider a distributed setting, where each agent (generators and loads) in the power network is only privy to their own local objectives and constraints but can communicate with their neighbours. A distributed online algorithm is proposed based on the modified primal-dual approach. The performance of the online algorithm is evaluated using the regret (static) function, which is the difference between the actual cost incurred by employing the proposed algorithm and the optimal fixed decision in hindsight. Since we deal with a constrained optimization problem, analogous to the notion of regret the accumulation of the constraint violation is also calculated at each step. We establish a sub-linear bound on the static regret and constraint violation under suitable assumptions on step-size and cost function. Finally, we use the standard IEEE-14 bus system to demonstrate the performance of our algorithm.