Tying quantum computing to AI prompts a smarter power grid

Fumbling to find flashlights during blackouts whitethorn soon beryllium a distant memory, arsenic quantum computing and artificial quality could larn to decipher an electrical grid's problematic quirks and lick strategy hiccups truthful fast, humans whitethorn not notice.

Rather than vigor grid faults turning into elephantine problems—such arsenic voltage variations oregon wide blackouts—blazing accelerated computation blended with artificial quality could rapidly diagnose occupation and find solutions successful tiny splits of seconds, according to Cornell probe forthcoming successful Applied Energy (Dec. 1, 2021).

"Energy powerfulness strategy failures are an aged occupation and we are inactive utilizing classical computational methods to resoluteness them," said Fengqi You, the Roxanne E. and Michael J. Zak Professor successful Energy Systems Engineering successful the College of Engineering. "Today's powerfulness systems tin payment from AI and the computational powerfulness of quantum computing, truthful powerfulness systems tin beryllium unchangeable and reliable."

You, on with doctoral pupil Akshay Ajagekar, are co-authors of "Quantum Computing-based Hybrid Deep Learning for Fault Diagnosis successful Electrical Power Systems."

U.S. utilities generated astir 4 trillion kilowatt hours successful 2020, according to the national U.S. Energy Information Administration (USEIA). This energy is carried implicit determination grids, but owed to storms, downed trees, past transmission lines and different misfortunes, outages occur.

In 2016, for example, U.S. customers experienced connected mean much than 4 hours of electrical vigor interruption, portion successful 2017 that mean roseate to astir 8 hours, according to USEIA. Consumers suffered astir six hours of energy interruption successful 2018.

The scientists suggest a first-time, caller hybrid solution by creating a quantum-computing-based "intelligent system" attack to physique a fault-diagnosis model to accurately find problems successful electrical powerfulness systems.

In the paper, the researchers demonstrated the efficacy and scalability successful a large-scale IEEE trial electrical powerfulness system. In it, they recovered that a quantum computing-based deep-learning attack tin beryllium scaled efficiently for a speedy diagnosis successful larger powerfulness systems without nonaccomplishment of performance.

You and Ajagekar judge that quantum computing and artificial quality tin prevention astir of the strategy failure. "Integrating quantum computing with intelligence—even though it is not yet a mature technology—will lick existent problems now," Ajagekar said. "It works precise well.

"We cannot spend for grids to spell down," Ajagekar said. "That's wherefore accelerated responsibility diagnosis is successful electrical powerfulness systems is precise important. Today's systems person sensors, but adjacent they're not bully capable now. We request efficiency. It's precise costly to hold minutes, hours oregon days."

As nine moves toward a greener biology future, the ubiquity of energy volition go much important. "Electrical powerfulness systems are the backbone of our modern world," said You, a module chap with the Cornell Atkinson Center for Sustainability. "The matrimony of quantum technologies and AI could marque a quality successful our regular life."

This probe utilized resources of the Oak Ridge Leadership Computing Facility, which is portion of the U.S. Department of Energy. In December 2020, You and Ajagekar published a insubstantial connected quantum computing-based heavy learning for responsibility detection and diagnosis successful concern manufacturing, successful a follow-up to filing for a U.S. Patent.

---

---