Power shift: U of T researcher applies AI to monitor city’s electrical grid

From indoor lighting to outdoor street lamps, our world is made brighter by artificial light. But the light that we perceive to be constant, actually fluctuates.

A University of Toronto computer scientist and researchers from the Technion-Israel Institute of Technology are studying electrical grids for cities, creating a camera that records the city's lights at a slower speed to get more accurate readings of changing voltages at particular locations. 

The hope is that their research could help utility companies monitor how voltage shifts and propagates throughout the grid, partly to regulate it and partly to monitor any instability – like a blackout. 

“Lights flicker because they are powered by alternating current, and by different phases of the electric grid,” says University of Toronto department of computer science Professor Kyros Kutulakos. “With a camera, remotely, we can begin to observe a big part of the city, and its electrical phases.”

Computational Imaging on the Electric Grid: Haifa, Israel by night captured in slow motion

Kutulakos and lead researchers Mark Sheinin and Professor Yoav Schechner of Technion-Israel Institute of Technology, are presenting their study Computational Imaging on the Electric Grid on July 22 at the Computer Vision and Pattern Recognition (CVPR) conference of the Institute of Electrical and Electronics Engineers (IEEE). 

Computer vision is a subfield of artificial intelligence, a burgeoning field that has put a spotlight on Toronto and U of T in recent months. The researcher’s methods merge other areas including optics, image processing and electrical grid engineering. 

To collect this data, the researchers needed to capture the lights’ flicker. Light emitted from light connected to the electricity grid is constantly changing, but because of the high speed of this effect, people do not sense this flickering.

“We couldn’t just capture a long exposure image, because then all the variations would be averaged out. On the other hand, we are also dealing with scenes at night – there isn’t enough light to do it,” says Kutulakos (below).

Last year, the researchers received Mitacs support for Sheinin to come to Toronto to help Kutulakos build their “ACam” – an alternating current camera – designed to operate and capture the grid’s light pluses.

“We needed a camera that allows us to control exactly when each pixel will record light,” says Kutulakos, who has worked on computational imaging techniques for many years. “In this case, it's always going to be synchronized with the alternating current. The camera shutter remains open for two, three seconds, but it's not going to record light continuously. In the end, we'll have enough light to give us a single image.”

Sheinin returned to Haifa, Israel, where he built a replica of the ACam before returning the original to Toronto.

Currently, their ability to capture the many flickers as a function of time, demonstrates their phase on the grid. Their database can also distinguish between the different types of light sources, from the sodium street lamps, to LEDs.

“The hope is that eventually we'll be able to get at much denser measurement on the state of the grid, and understand what the voltage is at that particular location, optically, without physically connecting to it,” says Kutulakos.

But this research isn’t limited to the outdoors. It could be applied to indoor spaces, computationally testing different types of artificial lighting.

“We're just starting to scratch the surface of this problem.”

The research was supported by the Taub Foundation, the Israel Science Foundation and the German Minerva Foundation, the Natural Sciences and Engineering Research Council of Canada, the Mitacs Canada-Israel Globalink Innovation Initiative and DARPA.