Graph Convolutional Neural Networks for Optimal Power Flow Locational Marginal Price

TL;DR

This paper introduces Chebyshev Graph Convolutional Neural Networks to predict real-time locational marginal prices in power grids, enhancing efficiency and scalability over traditional iterative methods.

Contribution

The paper presents a novel GCN-based approach for real-time OPF price prediction, addressing computational challenges and scalability issues in renewable-integrated power systems.

Findings

Achieves faster prediction of marginal prices

Demonstrates improved scalability and efficiency

Validates results through numerical benchmarking

Abstract

The real-time electricity market with the integration of renewable energies and electric vehicles have been receiving significant attention recently. So far most of the literature addresses the optimal power flow (OPF) problem in the real-time electricity market context by iterative methods. However, solving OPF problems in real-time is challenging due to the high computational complexity by the iterative methods. Motivated by this fact, in this paper, we propose a Chebyshev Graph Convolutional Neural Networks (ChebGCN) to improve the efficiency of integrating low-carbon energy sources into power grids and to address scalability and adaptivity of end-to-end existing OPF solutions. The proposed GCN method is capable to predict the optimal energy market marginal prices in real time. Numerical analysis is used to benchmark the results and validate the improvement.