AI Supercharging Battery Development for Electric Vehicles

Artificial intelligence is set to change recharging an EV and help improve other aspects of battery technology.

FREMONT, CA: From driving range to charging time to the vehicle's lifetime, battery performance can make or break the electric car experience. Artificial Intelligence (AI) has made it a reality to recharge an EV in the time it takes to stop at a gas station and improve other aspects of battery technology. For years, innovations in electric vehicle batteries have been hindered by a significant bottleneck: evaluation times. At each stage of the battery development procedure, the latest technologies must be tested for months to decide how long they will last. But, battery testing for electric vehicles surged forward with AI to cut testing times by about 15fold. Here is more to know.

A tea of researchers has developed a machine learning-based method that slashes these testing times by 98 percent. However, the group tested the battery charge speed method and found that it can be applied to several other parts of the battery development pipeline and even to non-energy technologies. In battery testing, there is a need to try many things because the performance will vary drastically. With AI, the researchers can rapidly identify the most promising methods and cut out a lot of unnecessary experiments. Researchers figured out how to greatly boost the testing process for extremely fast charging. This approach can be applied to many other problems holding back battery development for months or years.

Designing ultra-fast-charging batteries is a significant challenge, mainly because it is daunting to make them last. The faster charge's intensity puts a strain on the battery, which often causes it to fail early. To avoid damage to the battery pack, a component that accounts for a chunk of an electric car's cost, battery engineers must test a series of charging methods to find the ones that work best. The new research sought to streamline this process. At the outset, the team found that fast-charging streamlining amounted to several trial-and-error tests that is inefficient for humans but the perfect issue for a machine.