SafePowerGraph: Safety-aware Evaluation of Graph Neural Networks for Transmission Power Grids

TL;DR

SafePowerGraph addresses the lack of safety-oriented evaluation for graph neural networks in power-grid operations by delivering a simulator-agnostic framework that tests PF/OPF predictions under constrained perturbations. It integrates multiple PF/OPF simulators, constructs heterogeneous graph representations, and trains with a composite loss that enforces physics and safety constraints, including self-supervised terms. Key findings show Graph Attention Networks (GAT) consistently outperform alternatives and that self-supervised learning enhances robustness to perturbations like energy price shifts, while price variations significantly threaten safety margins on larger grids. The work provides a public leaderboard, dataset zoo, and model zoo, aiming to standardize robust, safe GNN research for real-world power systems and accelerate deployment of reliable ML-based grid analytics.

Abstract

Power grids are critical infrastructures of paramount importance to modern society and their rapid evolution and interconnections has heightened the complexity of power systems (PS) operations. Traditional methods for grid analysis struggle with the computational demands of large-scale RES and ES integration, prompting the adoption of machine learning (ML) techniques, particularly Graph Neural Networks (GNNs). GNNs have proven effective in solving the alternating current (AC) Power Flow (PF) and Optimal Power Flow (OPF) problems, crucial for operational planning. However, existing benchmarks and datasets completely ignore safety and robustness requirements in their evaluation and never consider realistic safety-critical scenarios that most impact the operations of the power grids. We present SafePowerGraph, the first simulator-agnostic, safety-oriented framework and benchmark for GNNs in PS operations. SafePowerGraph integrates multiple PF and OPF simulators and assesses GNN performance under diverse scenarios, including energy price variations and power line outages. Our extensive experiments underscore the importance of self-supervised learning and graph attention architectures for GNN robustness. We provide at https://github.com/yamizi/SafePowerGraph our open-source repository, a comprehensive leaderboard, a dataset and model zoo and expect our framework to standardize and advance research in the critical field of GNN for power systems.

Figures (12)

• Figure 1: Architecture of the SafePowerGraph framework.
• Figure 2: Evolution of energy prices, reported by the IEA iea2022.
• Figure 3: Embedding for a 9-bus grid. In purple the generators, blue the slack, orange the buses, green the loads, and red the lines.
• Figure 4: Transforming a homogeneous GNN to a heterogeneous GNN by combining message passing across multiple types. Here with two node types bus and load, and Sage layers.
• Figure 5: Impact of power grid perturbations on the robustness and safety of the OPF predictions.