Annoucement, slides and recordings can be found on the PSERC Webinar website

Abstract:

With increased uncertainties and rapidly changing operational conditions in power systems, existing stability control methods and operation paradigms have outstanding issues in terms of either speed, adaptiveness, or scalability. Recent years have seen notable progress in AI and learning-based control methods such as deep reinforcement learning (DRL) for solving challenging control and decision-making problems across many domains such as games, robotics and power systems. However, existing methods still have scalability, adaptability, and security issues. To address these challenges, an integrated framework based on the idea of Convergence of AI, Physics, Computing, and Control is developed. Based on this framework, scalable, physics-informed DRL algorithms and highperformance computational tools are developed to achieve efficient training of DRL agents for intelligent stability control for largescale power systems. The developed methods have been tested and demonstrated with large-scale power systems. Finally, this presentation will discuss the potential of this framework, when combined with new hardware and software platform, for transforming the grid operation and control from the control rooms to the grid edge.