Project Grant 2625248

The National Science Foundation Division of Electrical, Communications and Cyber Systems awarded the University of Denver $314,144 on August 15, 2026, for physics-grounded deep reinforcement learning research aimed at adaptive out-of-step protection in low-inertia power grids under the Engineering program (CFDA 47.041).

The project develops an artificial intelligence framework that enables protective relays to continuously adapt their behavior based on real-time grid conditions rather than relying on predetermined fixed parameters. The work combines physics-based analysis of grid dynamics with machine learning while maintaining transparency and verifiability of protection decisions for engineers. The research addresses a critical threat to grid reliability: out-of-step conditions where loss of synchronization can cause permanent equipment damage and trigger cascading blackouts. As inverter-based resources expand and system inertia declines, these conditions become more likely and exceed the capabilities of existing protective relays.

The project encompasses research and development activities, graduate student training through research participation, an NSF-funded undergraduate research program, and K-12 outreach in electrical engineering. Performance takes place in Denver, Colorado, with a period of performance extending through July 31, 2029. The award is classified as a Project Grant, an assistance type supporting investigator-initiated research rather than procurement contracting.