Real-Time HIL Validation of Heterogeneous Graph Neural Network for IBR-Dominated Distribution Grids

The increasing application of machine learning (ML) in power system research necessitates validation under realistic conditions, with real-time hardware-in-the-loop (HIL) simulations serving as a critical tool for integrating computational models with physical systems to enable comprehensive testing across diverse scenarios. This study introduces the HGNN-HIL framework for evaluating Graph Neural Network (GNN) agents, utilizing Hypersim to simulate IBR-dominated distribution grids and generate high-fidelity datasets. These data sets, transmitted through SCADA to an AWS cloud database, are used by a heterogeneous graph neural network (HGNN) for estimating the state of the power system and dynamic analysis. The framework captures complex grid interactions and generates critical data for HGNN training. Validated under varying load conditions, the HGNN demonstrates high accuracy, robustness, and stability in state prediction. This research highlights the potential of combining HIL simulations with advanced neural networks to enhance real-time power system operations, marking a significant step toward intelligent, adaptive control strategies for modern resilient grids.