Machine learning (ML) technology may be key to reducing the time needed to charge electric vehicles (EVs) and foster greater penetration of this mobility mode. U.S. Idaho National Laboratory (INL) researchers used ML techniques that incorporate charging data to generate unique charging protocols for superfast charging methods tailored to different types of EV batteries.

Speeding up the charging process with current technology can damage a battery. Inducing faster lithium ion migration from the cathode to the anode can result in lithium metal build-up, triggering early battery failureAnalyzing data at INL’s Battery Test Center. Source: INL and cathode cracking.

By inputting information about the condition of many lithium-ion batteries during their charging and discharging cycles, the ML program was trained to predict lifetimes and the ways that different battery designs would eventually fail. That data was fed back into the analysis to identify and optimize new protocols that were then tested on real batteries.

The results presented at the fall meeting of the American Chemical Society demonstrate that different types of EV batteries can charge to over 90% in 10 minutes by use of the newly devised protocols, and without lithium plating or cathode cracking.