A Lecture from Dr. Ying Zhang of Oklahoma State University titled: Trustworthy AI for Smart Distributed Energy Systems: Linking Physical and Data Intelligence

Abstract:

The rapid proliferation of distributed energy resources (DERs), smart meters, and intelligent sensors is reshaping modern power systems into highly dynamic, data-rich cyber-physical ecosystems. However, ensuring trustworthy, explainable, and resilient operation requires bridging the long-standing gap between physical intelligence—grounded in physics-based modeling, topology, and operational constraints—and data intelligence, derived from machine learning, statistical inference, and data analytics. This talk presents a unified framework and learning-enabled solutions that tightly couple these two forms of intelligence to enable next-generation situational awareness, control, and resilience in distributed grids, paving the way for new AI frontiers. Physical intelligence provides interpretability, feasibility, and safety guarantees, while data intelligence offers adaptability, predictive capability, and robustness to noisy and incomplete information. Their integration, achieved through physics-informed regression, graph-based models, and reinforcement learning, allows for consistent monitoring and decision-making on the grid edge and across the system. We will demonstrate that, by harmonizing domain physics with data-driven intelligence, the envisioned cognitive smart grid will evolve toward learning-enabled, self-healing, and trustworthy distributed energy systems capable of operating reliably under uncertainty and extreme events.