New machine learning model developed to prevent electric vehicle battery fires

One of the most important safety concerns for electric vehicles is keeping their batteries cool, as temperature spikes can have dangerous consequences.

New research led by a University of Arizona doctoral student proposes a way to predict and prevent temperature spikes in the lithium-ion batteries commonly used to power these vehicles.

The article, “Advancing Battery Safety,” led by Basab Goswami, a PhD student in the Faculty of Engineering, is published in the Energy Sources Journal.

Machine learning

With $599,808 in support from the Department of Defense's Competitive Research Incentive Program, Goswami and his advisor, aerospace and mechanical engineering professor and project principal investigator Vitaliy Yurkiv, developed a framework that uses multiphysics and machine learning models to detect, predict and identify overheating in lithium-ion batteries, known as thermal runaway.

Goswami said that in the future, this framework could be integrated into an electric vehicle's battery management system to prevent the battery from overheating, thereby protecting drivers and passengers.

“We need to move to green energy,” Goswami said, “but there are safety concerns about lithium-ion batteries.”

Thermal runaway can be extremely dangerous and difficult to predict.

“The temperature in a battery will increase exponentially and that will cause a fire,” Goswami said.

Domino effect

An electric vehicle battery pack is made up of tightly connected battery “cells.” Current electric vehicles can have more than 1,000 cells in each battery pack.

If thermal runaway occurs in one cell, neighboring cells are likely to heat up as well, creating a domino effect. If that happens, the entire electric vehicle battery could explode, Goswami said.

To avoid this, the researchers propose using thermal sensors – wrapped around the battery cells – that feed historical temperature data into a machine learning algorithm to predict future temperatures.

The algorithm predicts when and where an uncontrollable event is likely to occur.

“If we know the location of the hot spot (the start of thermal runaway), we can have solutions to shut down the battery before it reaches that critical stage,” Goswami said.

Yurkiv was impressed by the accuracy of Goswami’s algorithm. Before his research, machine learning models had not been used to predict thermal runaway.

“We didn’t expect machine learning to be so superior in predicting thermocouple temperature and hot spot locations so accurately,” Yurkiv said.

“No human would ever be able to do that.”

The research builds on a paper published by Goswami and Yurkiv in January, which explores using thermal imaging to predict runaway events, which would require heavy imaging equipment constantly taking pictures to examine them.

The solution identified by Goswami and Yurkiv in their latest paper is lighter and more cost-effective.

Goswami’s research comes at a significant moment in American automaking history. In July, the same month the study was released, the Biden administration announced a $1.7 billion investment in electric vehicle manufacturing in eight states.

In 2023, global sales of electric vehicles are expected to increase by 35% compared to 2022. Goswami said that as demand increases, Goswami said that safety measures are essential for the circulation of electric vehicles.

“Many people are still hesitant to adopt batteries because of various safety concerns,” he said. “For these technologies to become widely accepted, it is essential that the public is aware that ongoing research is actively addressing these critical safety issues.”