Learning-Based, Safety and Stability-Certified Microgrid Control

A neural-Lyapunov-barrier-enabled, physics informed-learning-based control method is devised to provide certificated safe and stable hierarchical control of microgrids. The main contributions include: 1) a neural hierarchical control framework for microgrids with provable safety and stability guarantees; 2) a control Lyapunov barrier function (CLBF) considering the fast dynamics of distributed energy resources, loads, and networks in microgrids; 3) a physics-informed learning approach for CLBF-based neural hierarchical control synthesis, which learns safety and stability certificates and control policy simultaneously without a verification module. Case studies demonstrate the effectiveness of the approach in provably certifying the stability and safety of microgrids equipped with hierarchical inverter control.