PNNL featured experts are available to provide scientific and technical expertise to the news media. To arrange an interview with a PNNL expert, contact the PNNL News & Media team. To search the entire database, visit Featured Experts.
Draguna L. Vrabie, PhD
Chief Scientist, Data Sciences and Machine Intelligence
Draguna L. Vrabie, PhD
Chief Scientist, Data Sciences and Machine Intelligence
Biography
Energy flexibility provides a more cost-effective, clean, and resilient energy system. A flexible energy system balances energy demand, supply, distribution, and storage resources such as the grid, renewables, batteries, and even electric vehicles. The key to making it work is rapidly sharing information and intelligently making decisions across many systems, including the power grid, residential homes, and commercial buildings. Accomplishing this delicate balance requires autonomous decision-making and control by millions of intelligent devices.
Vrabie’s research emphasizes the use of machine learning methods in the design of autonomous control systems.
“Intelligent control systems can monitor their environment, learn from past observations, make predictions about the results of future actions, and thus can optimize action plans in the future,” she said. “This approach to learning to make good decisions is similar to how humans make decisions.”
She uses deep reinforcement learning techniques to create reliable, sustainable, and cyber-secure intelligent energy systems. For example, in grid-interactive efficient buildings, Vrabie and her team developed physics-informed machine learning methods to accurately predict building response, to monitor the state of health of energy systems, and to detect the presence of malicious intruders. Such models could serve as realistic cyber decoys that persuade stealthy attackers to reveal their strategies.
In 2020, she received the Best Paper Award from the Energy Systems Technical Committee of the American Society of Mechanical Engineers Dynamic Systems and Control Virtual Conference. In 2017, she was named “Engineer of the Year” by the Women in Engineering section of the Institute of Electrical and Electronics Engineers (IEEE) in Richland, WA. Since 2019, she has served as a technical leader for PNNL’s Data Model Convergence Initiative, where she investigates algorithmic and computational methods for applications that integrate scientific modeling and simulation with data analytics and machine learning. Vrabie has co-authored two books, and she holds a doctorate in electrical engineering for her work in reinforcement learning.