PF$\Delta$: A Benchmark Dataset for Power Flow under Load, Generation, and Topology Variations
Researchers have introduced PF$\Delta$, a new benchmark dataset designed to evaluate machine learning models for power flow calculations. This dataset includes over 850,000 solved instances across various system sizes and contingency scenarios, aiming to address the computational bottlenecks in real-time grid operations. The paper also evaluates existing solvers and graph neural network-based methods, identifying challenges and future research directions for improving efficiency and accuracy in power system simulations. AI
IMPACT Provides a standardized benchmark for evaluating AI's potential to accelerate critical power grid simulations.