首页|Graph reinforcement learning for real-time dynamic reconfiguration and fault management in energy storage networks
Graph reinforcement learning for real-time dynamic reconfiguration and fault management in energy storage networks
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
Elsevier
With the booming development of renewable energy, the importance of energy storage systems in modern power systems has become increasingly prominent. Sufficient energy storage not only boosts the flexibility and stability of the power system but also facilitates the extensive utilization of renewable energy sources. This paper presents a Dynamic Reconfiguration Optimization (DRO) model. In the face of the challenges of energy storage systems in dynamic environments, this model, which combines graph neural networks and reinforcement learning, is designed to enhance the flexibility and fault management capabilities of energy storage systems. Empirical research on the IEEE 34 and 123 bus test networks reveals that, under various fault conditions and network scenarios, the DRO model can effectively improve energy supply and reduce voltage constraint violations. As a result, it successfully guarantees the economic viability and safety of the energy storage system. Compared with traditional methods, DRO shows remarkable decision-making accuracy and network reconfiguration capabilities, highlighting its effectiveness in managing complex energy storage equipment and formulating charging/discharging schedules. In conclusion, this study provides substantial technical support for constructing intelligent, efficient, and reliable energy storage systems, thus contributing to the sustainable development of the power industry.
Energy storageDynamic reconfigurationGraph neural networksFault managementReinforcement learningDISTRIBUTION-SYSTEMSSTRATEGY
NETL
NSTL
Elsevier
