Authors: Anna Varbella, Kenza Amara, Blazhe Gjorgiev, Giovanni Sansavini

Published on: February 05, 2024

Impact Score: 8.3

Arxiv code: Arxiv:2402.02827

Summary

• What is new: The first graph dataset (PowerGraph) modeling cascading failures in electric power grids specifically for training and explaining GNN models.
• Why this is important: Lack of public datasets of electrical power grids for GNN applications and the challenge in assessing vulnerability and identifying critical components in power grids due to scarcity of historical blackout data.
• What the research proposes: Developing PowerGraph, a dataset for training GNNs to detect cascading failures in power grids, using a physics-based model for generating diverse failure scenarios.
• Results: PowerGraph enables the development of better GNN models for various graph-level tasks and explainability, applicable across multiple domains.

Technical Details

Technological frameworks used: Graph Neural Networks (GNN)

Models used: Physics-based cascading failure model

Data used: Generated dataset modeling cascading failures in power grids

Potential Impact

Electric power grid management companies, GNN technology providers, industries relying on GNN for assessing system vulnerabilities.

Want to implement this idea in a business?

We have generated a startup concept here: GridIntelligence.