Optimal Power Flow Using Graph Neural Networks

  title={Optimal Power Flow Using Graph Neural Networks},
  author={Damian Owerko and Fernando Gama and Alejandro Ribeiro},
  journal={ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
Optimal power flow (OPF) is one of the most important optimization problems in the energy industry. In its simplest form, OPF attempts to find the optimal power that the generators within the grid have to produce to satisfy a given demand. Optimality is measured with respect to the cost that each generator incurs in producing this power. The OPF problem is non-convex due to the sinusoidal nature of electrical generation and thus is difficult to solve. Using small angle approximations leads to a… 

Figures and Tables from this paper

Unsupervised Optimal Power Flow Using Graph Neural Networks

A novel barrier method is proposed that is differentiable and works on initially infeasible points, and it is shown that the use of GNNs in this unsupervised learning context leads to solutions comparable to standard solvers while being computationally efficient and avoiding constraint violations most of the time.

Proximal Policy Optimization with Graph Neural Networks for Optimal Power Flow

A novel architecture based on the Proximal Policy Optimization algorithm with Graph Neural Networks to solve the Optimal Power Flow is proposed, which is to design an architecture that learns how to solves the optimization problem and that is at the same time able to generalize to unseen scenarios.

Power Flow Optimization with Graph Neural Networks

Several supervised GNN models are proposed to solve the power flow problem, using established GNN blocks from the literature, and the experimental results show that the GNNs are comparatively successful at generalizing to widely different topologies seen during training, but do not manage to generalize to unseen topologies and are not able to outperform an MLP on slight perturbations of the same energy system.

Solving AC Power Flow with Graph Neural Networks under Realistic Constraints

A model architecture on which unsupervised training is performed to learn a general solution of the AC power power formulation that is independent of the specific topologies and supply tasks used for training is presented.

Reduced Optimal Power Flow Using Graph Neural Network

A new method to reduce the number of constraints in the original OPF problem using a graph neural network (GNN) is presented, an innovative machine learning model that utilizes features from nodes, edges, and network topology to maximize its performance.

Learning to Solve AC Optimal Power Flow by Differentiating through Holomorphic Embeddings

This approach constitutes the first learning-based approach that successfully respects the full non-linear AC-OPF equations and reports a 12x increase in speed and a 40% increase in robustness compared to a traditional solver.

Learning to Optimize Power Distribution Grids using Sensitivity-Informed Deep Neural Networks

Numerical tests showcase that sensitivity-informed deep learning can enhance prediction accuracy in terms of mean square error (MSE) by 2-3 orders of magnitude at minimal computational overhead.

Graph Neural Networks for Voltage Stability Margins With Topology Flexibilities

High penetration of distributed energy resources (DERs) changes the flows in power grids causing thermal congestions which are managed by real-time corrective topology switching. It is crucial to

Learning to Solve the AC Optimal Power Flow via a Lagrangian Approach

This work uses deep neural networks to learn the dual variables of the ACOPF problem and proposes a Lagrangian-based approach that can reach more globally optimal solutions with significant computational speedup even when the training data consists of mostly suboptimal solutions.

Unsupervised Deep Learning for AC Optimal Power Flow via Lagrangian Duality

A deep neural network is developed to output a partial set of decision variables while the remaining variables are recovered by solving AC power flow equations and the fast decoupled power flow solver is adopted to further reduce the computational time.



Toward Distributed Energy Services: Decentralizing Optimal Power Flow With Machine Learning

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, providing a framework for Distribution System Operators to efficiently plan and operate the contributions of DERs to achieve Distributed Energy Services in distribution networks.

Machine Learning for AC Optimal Power Flow

This work presents two formulations of ACOPF as a machine learning problem: 1) an end-to-end prediction task where the optimal generator settings are predicted, and 2) a constraint predictiontask where the set of active constraints in the optimal solution are predicted.

History of Optimal Power Flow and Formulations

The purpose of this paper is to present a literature review of the AC Optimal Power Flow (ACOPF) problem and propose areas where the ACOPF could be improved. The ACOPF is at the heart of Independent

A power flow method suitable for solving OPF problems using genetic algorithms

The standard procedure in which one sets the generator-bus active power output and voltage magnitude has been replaced by an innovative approach where generator- bus voltage magnitude and voltage angle are scheduled.

Regression-based Inverter Control for Decentralized Optimal Power Flow and Voltage Regulation

A systematic and data-driven approach to determine reactive power inverter output as a function of local measurements in a manner that obtains near optimal results and allows for an efficient volt-VAR optimization (VVO) scheme.

Convex Relaxations of Optimal Power Flow Problems: An Illustrative Example

This paper demonstrates that physically based conditions cannot universally explain algorithm behavior and uses an example OPF problem with two equivalent formulations to illustrate relaxations from the Lasserre hierarchy for polynomial optimization and a related “mixed semidefinite/second-order cone programming” hierarchy.

Optimal Wireless Resource Allocation With Random Edge Graph Neural Networks

This work introduces the random edge graph neural network (REGNN), which performs convolutions over random graphs formed by the fading interference patterns in the wireless network, and presents an unsupervised model-free primal-dual learning algorithm to train the weights of the REGNN.

Lecture Notes on Optimal Power Flow (OPF)

These lecture notes cover the DC Optimal Power and ACoptimal Power Flow formulations, as well as the Economic Dispatch for Power Systems, and will include OPF formulations based on semidefinite programming, detailed derivation of Locational Marginal Prices, and other topics.

Stability Properties of Graph Neural Networks

This work proves that graph convolutions with integral Lipschitz filters, in combination with the frequency mixing effect of the corresponding nonlinearities, yields an architecture that is both stable to small changes in the underlying topology, and discriminative of information located at high frequencies.

MATPOWER: Steady-State Operations, Planning, and Analysis Tools for Power Systems Research and Education

The details of the network modeling and problem formulations used by MATPOWER, including its extensible OPF architecture, are presented, which are used internally to implement several extensions to the standard OPF problem, including piece-wise linear cost functions, dispatchable loads, generator capability curves, and branch angle difference limits.