Researchers aim to secure electric grid with artificial intelligence tools
July 22, 2024
Editor’s note: A version of this article originally appeared on Oak Ridge National Laboratory’s website.
UNIVERSITY PARK, Pa. — Penn State is among the institutions partnering with the Department of Energy’s Oak Ridge National Laboratory (ORNL) to launch a project to develop an innovative suite of tools that will employ machine learning algorithms for more effective cybersecurity analysis of the U.S. power grid.
Minghui Zhu, professor of electrical engineering in the Penn State College of Engineering, and Peng Liu, Raymond G. Tronzo, MD Professor of Cybersecurity in the Penn State College of Information Sciences and Technology, are co-PIs of the project. Zhu will focus on threat mitigation, while Liu will lead the task of vulnerability analysis.
Distributed energy resource (DER) systems, which can range from solar panels to electric vehicles to demand response programs, are reshaping traditional grid operations.
This evolution introduces new challenges for cybersecurity. The reliance on information and communication technologies to facilitate connectivity exposes utility systems to cyber threats, which can be effectively confronted using tools powered by artificial intelligence (AI), according to the researchers.
The new tool suite, known as AI-PhyX, streamlines collection and analysis of data to comprehensively tackle all facets of cyber resilience, including vulnerability analysis, attack detection, threat mitigation and system recovery. AI helps convert data into actionable information that enables system operators to make better decisions. Once the tool suite is developed, it will be demonstrated using system data provided by utility partners, such as EPB of Chattanooga, aiming for improved utility acceptance.
Researchers are developing a workflow that integrates diverse cybersecurity applications into one platform to streamline the training and operation of the system. The resulting tool suite helps solve challenges in data management that often lead to fragmented analysis and hamper deployment.
As part of Penn State’s contributions to the project, Liu will develop new AI-powered techniques for identifying unknown DER-system-specific vulnerabilities associated with the interactions between various components across the operational technology network/informational technology network interface.
“These new techniques and tools will enable the DER industry to gain better awareness and more comprehensive understanding about the security vulnerabilities associated with real world DER systems,” Liu said.
Zhu will develop physics-informed machine learning techniques to mitigate cyberattacks on DER systems. In particular, he will leverage physics-informed reinforcement learning to train new cyber defenses, which adaptively and autonomously deploy moving target defense techniques to minimize the damage caused by cyberattacks on DER systems.
“These new learning-based defenses will enhance optimized and automated defense capabilities of DER systems and maintain their critical functionality after security breaches,” Zhu said.
DER systems hold immense promise and could deliver benefits such as improved grid reliability, reduced electricity costs and decreased greenhouse gas emissions.
Development of the new tool suite is funded by the DOE Office of Cybersecurity, Energy Security and Emergency Response. At ORNL, Jamie Lian and Teja Kuruganti are the lead and co-lead of the project, respectively. The DOE’s National Renewable Energy Laboratory, the University of Connecticut and Siemens Corp are also contributing to the project as partner institutions.