Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SafePowerGraph-HIL: Real-Time HIL Validation of Heterogeneous GNNs for Bridging Sim-to-Real Gap in Power Grids

Aoxiang Ma, Salah Ghamizi, Jun Cao, Pedro Rodriguez

TL;DR

The paper addresses the sim-to-real gap in power-system ML validation by integrating real-time hardware-in-the-loop testing with a heterogeneous graph neural network ($HGNN$) for state estimation and dynamic analysis. It introduces the SafePowerGraph-HIL framework, combining real-time high-fidelity data generation in Hypersim, SCADA-to-AWS data transfer, and HGNN-based modeling of PF/OPF problems. Empirical results on the WSCC 9-Bus show strong synthetic-domain performance but notable domain shift when confronted with real-time data, which is mitigated by fine-tuning on hardware-generated data, demonstrating 13% and 75% reductions in error for buses and slack-state predictions respectively. The approach offers a robust, real-time pathway for monitoring and adaptive control in evolving grids, with potential to accelerate deployment of intelligent grid-management strategies while bridging the gap between simulation and real-world operation.

Abstract

As machine learning (ML) techniques gain prominence in power system research, validating these methods' effectiveness under real-world conditions requires real-time hardware-in-the-loop (HIL) simulations. HIL simulation platforms enable the integration of computational models with physical devices, allowing rigorous testing across diverse scenarios critical to system resilience and reliability. In this study, we develop a SafePowerGraph-HIL framework that utilizes HIL simulations on the IEEE 9-bus system, modeled in Hypersim, to generate high-fidelity data, which is then transmitted in real-time via SCADA to an AWS cloud database before being input into a Heterogeneous Graph Neural Network (HGNN) model designed for power system state estimation and dynamic analysis. By leveraging Hypersim's capabilities, we simulate complex grid interactions, providing a robust dataset that captures critical parameters for HGNN training. The trained HGNN is subsequently validated using newly generated data under varied system conditions, demonstrating accuracy and robustness in predicting power system states. The results underscore the potential of integrating HIL with advanced neural network architectures to enhance the real-time operational capabilities of power systems. This approach represents a significant advancement toward the development of intelligent, adaptive control strategies that support the robustness and resilience of evolving power grids.

SafePowerGraph-HIL: Real-Time HIL Validation of Heterogeneous GNNs for Bridging Sim-to-Real Gap in Power Grids

TL;DR

The paper addresses the sim-to-real gap in power-system ML validation by integrating real-time hardware-in-the-loop testing with a heterogeneous graph neural network () for state estimation and dynamic analysis. It introduces the SafePowerGraph-HIL framework, combining real-time high-fidelity data generation in Hypersim, SCADA-to-AWS data transfer, and HGNN-based modeling of PF/OPF problems. Empirical results on the WSCC 9-Bus show strong synthetic-domain performance but notable domain shift when confronted with real-time data, which is mitigated by fine-tuning on hardware-generated data, demonstrating 13% and 75% reductions in error for buses and slack-state predictions respectively. The approach offers a robust, real-time pathway for monitoring and adaptive control in evolving grids, with potential to accelerate deployment of intelligent grid-management strategies while bridging the gap between simulation and real-world operation.

Abstract

As machine learning (ML) techniques gain prominence in power system research, validating these methods' effectiveness under real-world conditions requires real-time hardware-in-the-loop (HIL) simulations. HIL simulation platforms enable the integration of computational models with physical devices, allowing rigorous testing across diverse scenarios critical to system resilience and reliability. In this study, we develop a SafePowerGraph-HIL framework that utilizes HIL simulations on the IEEE 9-bus system, modeled in Hypersim, to generate high-fidelity data, which is then transmitted in real-time via SCADA to an AWS cloud database before being input into a Heterogeneous Graph Neural Network (HGNN) model designed for power system state estimation and dynamic analysis. By leveraging Hypersim's capabilities, we simulate complex grid interactions, providing a robust dataset that captures critical parameters for HGNN training. The trained HGNN is subsequently validated using newly generated data under varied system conditions, demonstrating accuracy and robustness in predicting power system states. The results underscore the potential of integrating HIL with advanced neural network architectures to enhance the real-time operational capabilities of power systems. This approach represents a significant advancement toward the development of intelligent, adaptive control strategies that support the robustness and resilience of evolving power grids.
Paper Structure (23 sections, 6 equations, 5 figures)

This paper contains 23 sections, 6 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Our Approach
  3. SafePowerGraph-HIL Verification Framework Overview
  4. Real-Time High-Fidelity Data Generation
  5. Pre-Trained GNN Validation
  6. SafePowerGraph Framework
  7. Dataset generation
  8. Heterogeneous Graph Neural Networks (HGNN) Training Process
  9. Graph Neural Network Training
  10. Performance evaluation.
  11. Real-Time Hardware-In-the-Loop Simulation Module
  12. Real-time hardware simulator
  13. SCADA
  14. AWS cloud database
  15. Empirical Study
  16. ...and 8 more sections

Figures (5)

  • Figure 1: SafePowerGraph-HIL Verification Framework
  • Figure 2: WSCC 9-Bus System from HYPERSIM
  • Figure 3: Our Hardware in the loop infrastructure: Left: Hardware simulator, middle: SCADA interface, right: Real-Time database
  • Figure 4: Validation Loss for Bus Voltage and Angle
  • Figure 5: Validation Loss for Slack Active Power (P) and Reactive Power (Q)