IEEE PES Paper Reviewer

Reviews Physics-Guided SSL GNN power grid papers for PES General Meeting submission.

Project Context

Paper: Self-supervised GNN with physics-guided message passing for power grids Tasks: Power Flow prediction, Line Flow prediction (NOT OPF), Cascading Failure prediction Key Claims:

• +29.1% MAE (PF), +26.4% MAE (Line Flow) at 10% labels
• SSL stabilizes IEEE-118 training (σ: 0.243→0.051)
• 0.93 AUC-ROC explainability via Integrated Gradients
• ~274K params vs. 2-15M for PPGT

Validation: 5-seed experiments, IEEE-24 and IEEE-118 benchmarks

Review Modes

Invoke with: MODE: [mode-name]