Electrical Engineering and Computer Science Ph.D. Student wins Best Graduate Student Poster Award at International Power Systems Conference
Ali Bohluli Oskouei, a doctoral student and research assistant in electrical engineering and computer science at Syracuse University’s College of Engineering and Computer Science, won the Best Graduate Student Poster Award at the 19th International Conference on Probabilistic Methods Applied to Power Systems (PMAPS 2026).
The conference, sponsored by the IEEE Power & Energy Society, was held Sept. 21-24 in Salt Lake City, Utah.
Bohluli Oskouei presented his team’s work as both an oral research paper and a student poster. The paper was featured Sept. 22 in the “Data Center Grid Integration and Operational Flexibility” technical session.
The study, “A Meteorological Data-Aware Dynamic Load Model for Near-Real-Time Data Center Operations and Forecasting,” addresses two closely related problems: reconstructing missing or interrupted electricity demand measurements and forecasting demand up to 48 hours ahead using weather data and demand history.
The research team developed an artificial intelligence-based approach that combines historical data center electricity consumption patterns with weather information to better understand and predict changes in power demand. In tests using simulated data center operations, the approach achieved an average prediction error of just 0.56% when reconstructing missing measurements and approximately 7% when forecasting demand one and two days ahead, demonstrating its potential to support more accurate grid monitoring, planning, and reliability assessments.
“Winning this award means a lot to me, not only because it recognizes the effort we put into our research, but also because it reminds me why I chose this field in the first place,” said Bohluli Oskouei. “It motivates me to keep learning and contributing to the future of power systems. I’m truly grateful to my professors and Syracuse University for their support in reaching this milestone.”
Growing demand on the grid
The research comes as electricity demand in the United States is rising again after roughly two decades of flat growth. Artificial intelligence data centers account for much of the increase.
A single data center campus can now request hundreds of megawatts on timelines the grid was never designed to meet, while electrification adds pressure across other sectors. These loads are also unusually dynamic, with demand that shifts alongside computing activity and with cooling needs that follow the weather.
Grid operators and planners therefore need timely, dependable information about how such loads behave. That information helps them study grid stability, predict future demand and stay prepared when measurements are delayed or lost.
The results point to applications in grid monitoring, planning, simulation and digital twins.
Cross-disciplinary collaboration
The award-winning study grew out of a partnership between Sara Eftekharnejad, an associate professor of electrical engineering and computer science who directs the Power Systems Research Laboratory, and Chilukuri K. Mohan, a professor of electrical engineering and computer science. The collaboration pairs the lab’s expertise in power system modeling and reliability with Mohan’s work in machine learning and artificial intelligence.
The paper was coauthored by Bohluli Oskouei, Eftekharnejad and Mohan.