[Paper Review] Data-Driven Decentralized Optimal Power Flow.

Roel Dobbe, Oscar Sondermeijer+4 more|arXiv (Cornell University)|Jun 14, 2018

Optimal Power Flow Distribution32 references17 citations

TL;DR

This paper proposes a data-driven, decentralized optimal power flow (OPF) method that enables multiple distributed energy resources (DERs) to collaboratively approximate a centralized OPF solution using only local information. By leveraging a rate distortion framework, the approach learns local control policies that collectively achieve near-optimal performance while satisfying system constraints, with minimal communication requirements and a scalable framework for identifying optimal communication topologies.

ABSTRACT

The implementation of optimal power flow (OPF) methods to perform voltage and power flow regulation in electric networks is generally believed to require communication. We consider distribution systems with multiple controllable Distributed Energy Resources (DERs) and present a data-driven approach to learn control policies for each DER to reconstruct and mimic the solution to a centralized OPF problem from solely locally available information. Collectively, all local controllers closely match the centralized OPF solution, providing near-optimal performance and satisfaction of system constraints. A rate distortion framework facilitates the analysis of how well the resulting fully decentralized control policies are able to reconstruct the OPF solution. Our methodology provides a natural extension to decide what buses a DER should communicate with to improve the reconstruction of its individual policy. The method is applied on both single- and three-phase test feeder networks using data from real loads and distributed generators. It provides a framework for Distribution System Operators to efficiently plan and operate the contributions of DERs to active distribution networks.

Motivation & Objective

• To address the challenge of implementing optimal power flow (OPF) in distribution networks without relying on centralized coordination or extensive communication.
• To enable multiple distributed energy resources (DERs) to collectively approximate a centralized OPF solution using only locally available data.
• To develop a decentralized control policy learning framework that ensures system constraints are satisfied while achieving near-optimal performance.
• To provide a systematic method for determining which buses a DER should communicate with to improve reconstruction of the centralized OPF solution.
• To offer a practical framework for distribution system operators to plan and operate DER contributions in active distribution networks.

Proposed method

• Leverages a data-driven approach to train local control policies for each DER using historical or simulated operational data from the distribution network.
• Employs a rate distortion framework to quantify and optimize the trade-off between communication cost and reconstruction accuracy of the centralized OPF solution.
• Uses local measurements and limited inter-DER communication to learn control policies that collectively mimic the behavior of a centralized OPF solver.
• Applies a decentralized optimization framework where each DER learns its policy independently, yet the collective behavior converges to the centralized solution.
• Introduces a communication topology selection mechanism based on the rate distortion analysis to identify the most informative communication links for each DER.
• Validates the method on both single- and three-phase distribution test feeders using real-world load and DER data.

Experimental results

Research questions

• RQ1 Can decentralized DER control policies be learned from local data alone to closely approximate a centralized OPF solution?
• RQ2 How does the rate distortion framework enable the design of communication-efficient decentralized control policies?
• RQ3 What communication topology between DERs maximizes the accuracy of reconstructing the centralized OPF solution under limited communication?
• RQ4 To what extent can the decentralized policy achieve near-optimal performance while satisfying system constraints?
• RQ5 How does the method scale across different distribution network configurations, including single- and three-phase systems?

Key findings

• The decentralized control policies collectively achieve near-optimal performance that closely matches the centralized OPF solution in terms of power flow and voltage regulation.
• The rate distortion framework successfully identifies communication links that significantly improve policy reconstruction accuracy with minimal added communication overhead.
• The method maintains system constraints such as voltage limits and thermal line flows across all tested single- and three-phase test feeders.
• The approach enables DERs to operate effectively with minimal coordination, relying only on local measurements and a small set of strategic communications.
• The framework provides a scalable and practical solution for distribution system operators to integrate and manage DER contributions in active distribution networks.
• Empirical results on real load and generator data confirm the robustness and practical viability of the proposed data-driven decentralized OPF approach.