Physics-Informed Heterogeneous Graph Neural Networks for DC Blocker Placement

TL;DR

This work tackles the challenge of sparsely placing DC-current blockers to mitigate geomagnetically induced currents in power grids. It merges a heterogeneous graph neural network (HGNN) with a physics-informed neural network (PINN) to predict blocker locations while enforcing AC/DC power-flow physics via a surrogate solver, addressing the nonconvex MINLP nature of blocker placement. The PIHGNN framework demonstrates improved predictive accuracy and reliability over purely data-driven baselines and offers substantial speed advantages over traditional heuristics, with promising generalization across networks. By integrating graph-structured power-network representations and physics-based constraints, the approach enables scalable, practical GIC mitigation for resilient grid operation under geomagnetic disturbances.

Abstract

The threat of geomagnetic disturbances (GMDs) to the reliable operation of the bulk energy system has spurred the development of effective strategies for mitigating their impacts. One such approach involves placing transformer neutral blocking devices, which interrupt the path of geomagnetically induced currents (GICs) to limit their impact. The high cost of these devices and the sparsity of transformers that experience high GICs during GMD events, however, calls for a sparse placement strategy that involves high computational cost. To address this challenge, we developed a physics-informed heterogeneous graph neural network (PIHGNN) for solving the graph-based dc-blocker placement problem. Our approach combines a heterogeneous graph neural network (HGNN) with a physics-informed neural network (PINN) to capture the diverse types of nodes and edges in ac/dc networks and incorporates the physical laws of the power grid. We train the PIHGNN model using a surrogate power flow model and validate it using case studies. Results demonstrate that PIHGNN can effectively and efficiently support the deployment of GIC dc-current blockers, ensuring the continued supply of electricity to meet societal demands. Our approach has the potential to contribute to the development of more reliable and resilient power grids capable of withstanding the growing threat that GMDs pose.

Figures (6)

• Figure 1: An abstract view of a heterogeneous graph for power grid (B4GIC).
• Figure 2: PIHGNN Framework
• Figure 3: Evaluation of predictions. The heuristic solver provides the label $z$ (highlighted in yellow), and the GNN-based prediction $\hat{z}$ (highlighted in green) is evaluated by the evaluator $g$. The evaluator $g$ is a surrogate MLD model that takes the predictions from the heuristic solver and the GNN model as inputs.
• Figure 4: Training PIHGNN model on EPRI21
• Figure 5: A surrogate evaluator for EPRI21

Theorems & Definitions (1)

• Definition 1: Heterogeneous Graph