Physics-Informed Trustworthy Learning Towards Robust Situation-Aware Distribution Grids

Electrical and Computer Engineering Graduate Seminar:
Modern power systems face challenges due to increased complexity and volatility from integrating renewable and distributed energy resources (DERs). Effective grid management requires real-time situational awareness from diverse sensor data streams to support monitoring and decision-making. This talk introduces a new AI-based technique for power flow calculation in complex grids with many DERs, combining power engineering knowledge with data-driven methods. Unlike traditional “black-box” machine learning, this approach ensures faster, more accurate, and robust results, even under outliers or cyberattacks. The proposed framework has the potential to enhance grid control and optimization applications. Future research and collaboration opportunities will also be discussed.