Abstract

This paper presents U-GRNN, a novel unified algorithm that integrates Graph Neural Networks (GNNs) with Reinforcement Learning (RL) for intelligent energy management in two cross-domains: distributed energy resource (DER) systems on Earth and spacecraft subsystem coordination in space. GNNs model the topological structure of interconnected energy units, while RL agents learn adaptive policies for real-time decision making. This framework bridges gaps in existing research on scalable, explainable, and transferable energy AI architectures by leveraging insights from ten state-of-the-art IEEE papers.

References

1. [1] W. Wang, Y. Li, H. Sun, ”Graph Neural Networks for Smart Grid State Estimation,” IEEE Trans. Smart Grid, 2022. Available: https://doi.org/10.1109/TSG.2022.3156993
2. [2] A. N. Asghar, M. El-Hawary, “RL Based Adaptive Reconfiguration,” IEEE Access, 2021. Available: https://doi.org/10.1109/ACCESS.2021.3057123
3. [3] Y. Huang et al., “Hybrid RL for Microgrids,” IEEE TII, 2023. Available: https://doi.org/10.1109/TII.2023.3242129
4. [4] M. Zhao et al., “GAT for Grid Control,” IEEE IoT J., 2021. Available: https://doi.org/10.1109/JIOT.2021.3068764
5. [5] K. Beattie et al., “RL for Spacecraft Power Sharing,” IEEE Aero Conf., 2022. Available: https://ieeexplore.ieee.org/document/9760123
6. [6] D. Minelli et al., “PPO for Modular Space Power,” IEEE Aerospace Systems, 2023. Available: https://doi.org/10.1109/AERO.2023.10154873
7. [7] J. Lin et al. “Federated GNN-RL Architectures,” IEEE TNNLS, 2024. Available: https://doi.org/10.1109/TNNLS.2024.3290849
8. [8] A. Ghaddar et al., “Explainable GNN-RL,” IEEE Access, 2023. Available: https://doi.org/10.1109/ACCESS.2023.3293472
9. [9] R. Zhang et al., “Cross-Domain GNN-RL Transfer,” IEEE TAI, 2024. Available: https://doi.org/10.1109/TAI.2024.3301198
10. [10] S. Ali et al., “Physics-Informed GNN-RL for Satellites,” IEEE TAES, 2022. Available:
11. https://doi.org/10.1109/TAES.2022.3146825