Keeping the lights on.
AI for grids under attack, under strain, and under weather they were not built for.
GRAIL is a research lab in the Department of Electrical & Computer Engineering at The University of Tulsa, directed by Suman Rath. We build AI that is small enough to run on real grid hardware, honest enough to work on noisy and corrupted data, and specific enough to be useful to the people who actually operate power systems.
Our work sits at the intersection of power systems, machine learning, and cybersecurity: digital twins of real distribution networks, reinforcement learning agents that respond to attacks in real time, intrusion detection for microgrids, and models compressed to run on edge hardware. We started in August 2025. Everything on this site is something we have built, tested, published, or actively underway.
GRAIL grew out of the TU Power Lab in Rayzor Hall — built and led by Kaveh Ashenayi and Douglas Jussaume — which has trained Tulsa power engineers for decades, including many now at American Electric Power and Public Service Company of Oklahoma. GRAIL takes up that lineage and points it at a newer question: what happens to a grid under attack, under extreme weather, and under loads it was never planned for.
- Resilient grid control
- Grid cybersecurity
- Digital twins
- Reinforcement learning
- Edge AI
- Energy for AI & data centers
- AI for engineering education
What we work on
all research →Drone-mounted grid monitoring
A modular AI module that mounts on any commercial drone and classifies power line condition onboard, in milliseconds. Field-tested on a de-energized section of the Public Service Company of Oklahoma network.
Tulsa grid digital twin
A virtual replica of the Tulsa distribution network built entirely from open-source information — used to study how tornadoes, ice storms, and cyberattacks cascade through a real city's grid.
Cyber-resilient control
Unsupervised intrusion detection and reinforcement learning defenders that keep microgrid voltage and frequency stable while an attack is still in progress — not after.
The lab so far
News
| Sep 2026 | Shubham Raj Gupta's paper comparing machine learning and graph-based models for lithium-ion battery state-of-charge estimation is accepted at IEEE UEMCON 2026. |
| Aug 2026 | The IEEE PES Task Force report on resilient and secure large-scale Energy Internet systems, co-authored by S M Zia Ur Rashid and Suman, is posted on arXiv. |
| Aug 2026 | S M Zia Ur Rashid's paper on the double dip attack against cashback reward engines is published in Computers & Security. |
| Aug 2026 | Cedric Ogiesoba-Eguakun's paper on LLM-guided interactive simulation for power system dynamics education is accepted at IEEE UEMCON 2026. |
| Aug 2026 | Our paper on agentic AI for microgrid operational data integrity, with colleagues at Oak Ridge National Laboratory, is accepted at IEEE Resilience Week 2026. |
| Aug 2026 | Andrew Vidana and Matthew Reynolds were selected for NSF I-Corps for Fall 2026, for their AI-enabled grid simulator project. |
| Aug 2026 | Our work on machine learning assisted measurement denoising and cyber-resilient protection for grid-connected inverters is accepted at IEEE PEDES 2026, IIT Guwahati, in December. |
| Jul 2026 | Suman was selected for the NSF I-Corps regional program through the Southwest Hub, to develop and commercialize our lightweight cyberattack detection tool for macro and micro power systems. |
| 2026 | Suman is serving as Guest Editor for the Electronics special issue on Cybersecurity for Smart Grids: Latest Advances and Prospects (2026–2027). |
| May 2026 | GRAIL comes alive. Our group takes its name: the Grid Resilience and Artificial Intelligence Laboratory. |
| May 2026 | Two papers appear in IET Cyber-Physical Systems: Theory & Applications — on decentralized reinforcement learning for cyberattack response, and on ensemble federated learning for microgrid situational awareness. |
| May 2026 | Our senior design team's drone-mounted monitoring module was tested on a de-energized section of the Public Service Company of Oklahoma grid. Gwenith Madhan and Wyatt Fulton graduated and joined AEP and ONEOK. |
| Apr 2026 | Our framework for teaching power system dynamics with AI and simulation, led by Douglas Jussaume, is posted on arXiv. |
| Spring 2026 | Suman taught the inaugural offering of ECE 4863/6863 — AI in Power and Signal Systems, a new cross-listed course designed from scratch for 19 students. |
| Jan 2026 | Our survey on Federated Learning for Smart Grid: Applications and Potential Vulnerabilities is published in ACM Transactions on Cyber-Physical Systems. |
| Dec 2025 | Book chapter on The Security of Reinforcement Learning Systems in the Electric Grid Domain published by Wiley. |
| Aug 2025 | Suman joins The University of Tulsa as a tenure-track Assistant Professor in Electrical & Computer Engineering. GRAIL begins. |
| Jul 2025 | Invited talk to the IEEE PES Task Force on Resilient and Secure Large-Scale Energy Systems on reinforcement learning for resilient microgrid operations under unknown attack vectors. |