Study Presents AI-Inspired Solution for Assessing Energy Infrastructure Risks

The ISO New England control room in Holyoke, Mass, is responsible for monitoring and directing the flow of electricity across over 8,600 miles of transmission lines and more than 300 power plants—ensuring grid reliability for 14 million residents by balancing supply and demand. (Photo courtesy of ISO-NE)

"Advanced analytical tools like these are becoming indispensable as the electric grid grows increasingly complex, dynamic, and uncertain," said Junbo Zhao, the Todd M. Cook and Elizabeth Donohoe Cook Associate Professor of Engineering at Dartmouth and corresponding author on the study. "This work shows what's possible when researchers and industry partners come together to create practical, real-world solutions that improve reliability and keep the lights on for communities."

Recently published in Nature Communications, the paper, "Computationally efficient tail distribution-aware large-scale power system overloading risk assessment," was selected as a featured article in the "Engineering and Infrastructure" focus area.

"As the New England power grid continues faces increasing uncertainty arising from weather‑dependent generation and demand, it is critical to deepen our understanding of operational risk, particularly transmissions security under extreme weather conditions," said Tongxin Zheng, chief technologist for ISO New England's Advanced Technology Solutions group.

The new method significantly reduces the computational burden, as compared to traditional risk assessments, while improving accuracy in the 'tail' of risk distributions. Most planning tools look at what happens most of the time. Tail distribution looks at what happens at the edges, such as during severe heat waves or cold snaps that occur only occasionally but put the most strain on the power system.

Additional authors on the study are Xiaochuan Luo, Mingguo Hong, and Slava Maslennikov with ISO New England, and Professor Zhao's previous students Bendong Tan and Ketian Ye from the University of Connecticut.