How artificial intelligence can change electric vehicle charging

New AI tools can provide utilities with real-time data that will make the power grid and electric vehicle charging more reliable, although to a very miniature extent test by the University of Michigan Transportation Research Institute (UMTRI) and the startup Utilidata.

Scientists are using artificial intelligence to analyze electric vehicle charging behavior, hoping these insights will improve drivers’ experiences and facilitate utilities prepare for a surge in electricity demand. So far, they have found that charging electric vehicles can draw energy unevenly and result in poorer power quality, which can cause wear and tear on charging equipment.

These fundamental problems waste energy and can lead to failures of electric vehicle chargers, which have become a nightmare for drivers. So the ability to immediately detect and even predict AI problems could be a game-changer. The authors write that artificial intelligence models can give utilities information about how charging may affect the power grid. They can also advise drivers on where and when to charge, and facilitate electric vehicle charging companies better maintain their equipment.

UMTRI initially contacted Utilidata about this pilot study to facilitate develop a larger research project on the same topics. UMTRI says it is already working with the North American Electric Reliability Council to address preliminary findings.

For this study, researchers installed electricity meter adapters equipped with Utilidata’s Karman AI platform at six electric vehicle charging stations at the University of Michigan. Karman analyzed voltage, current, power and other dynamics between March and June last year. The study authors also installed devices in the vehicles of 10 drivers who frequently visit the university campus to monitor their charging habits.

While this project is still in its early stages, researchers hope it will facilitate people prepare for the challenges of electrifying their vehicle fleets. The U.S. already has aging power grids stretch to meet the growing electricity demand from AI data centers, cryptocurrency mining and neat energy technologies. However, compared to the data center, it is more hard for utilities to predict when and where EVs will be connected to the grid.

Utilities must deal with this unpredictability without real-time data to facilitate them adapt. These blind spots are becoming an increasing problem in “edge of the mesh”, where customers are increasingly connecting their own devices, such as electric vehicle batteries and solar panels, to the grid.

“Artificial intelligence has a big role to play at the edge,” says Siobhan Powell, a postdoctoral researcher at ETH Zürich who was not involved in the study. “It didn’t used to be like that, right? There wasn’t much interesting going on, and now that we have a chance to take control, knowing what’s going on provides more opportunities and more value.

One problem that researchers noticed in this study was short cycling and inconsistent power draw from vehicles that stopped and started charging even after the battery was fully charged. Not only does this burn energy inefficiently, but it can also overheat wires and transformers. They also found that charging electric vehicles degrades power quality when the electricity deviates from ideal voltage and frequency ranges. Flicker is a telltale sign of poor power quality, which can also cause greater wear and tear on your equipment.

“I think the most important takeaway is that we have confirmed that there is a lot of EV behavior that no one knows about – car owners, network operators, charger OEMs,” Utilidata vice president of product solutions, says Yingchen Zhang. “So there is a huge need to make all this data available.”

The study’s authors are careful to say that places with high levels of unmanaged EV charging could see a greater impact on the power grid. They say that in a worst-case scenario, energy supplies to other customers could be affected. However, Zhang is quick to say that the risk of a power outage as a result is very low.

“It’s good to know exactly how these new charges affect voltage and local power quality issues, but I wouldn’t jump right into blackouts,” Powell says, because there are many steps utilities can take to prevent failures. Again, this is a very small study of unpredictable charging behavior, so it’s too early to make sweeping statements about the impact of these early findings on the wider network.

Both Powell and Zhang want to avoid creating unnecessary anxiety about the impact that EV charging could have on the grid – especially as EV adoption faces partisan attacks. “A lot of concern is because people don’t know the actual behavior of electric vehicles,” Zhang says. “So actually disclosing that information will alleviate a lot of those concerns.”

The rise of artificial intelligence has also raised concerns that increasingly energy-hungry data centers are straining the network. Zhang says his company is also thinking about this and is using specially designed chips from Nvidia to use less power than more generic AI chips. And using machine learning in this way to analyze data is generally much less energy-intensive than generative AI models that spit out text and images.

It all comes down to preparation as the key to strengthening the energy grid against novel technologies that are changing the way we live, work and move. Fleets of electric vehicle batteries can even facilitate strengthen the grid by acting as virtual power plants, supplying energy to the grid when it is needed. Automakers are already testing this, in part to make electric vehicles more affordable for customers. “We need electric vehicles. We need this transition. There are also things we need to do to prepare the grid, but we can do it,” Powell says.