AI Can ‘Hear’ When a Lithium Battery Is About to Catch Fire

Researchers at the National Institute of Standards and Technology (NIST) have developed a way to use sound to detect when lithium-ion batteries are about to catch fire. The NIST team included Wai Cheong "Andy" Tam and Anthony Putorti.

Lithium-ion batteries are all around us: They can be in your phone, in your laptop, in an e-bike or an electric car. There may be several within arm's reach and hundreds of them in your building.

These batteries are popular because they can store a lot of energy in a small space. That quality makes them useful, but also brings danger. If a lithium-ion battery gets too hot or is damaged, it can catch fire or even explode. And the risk of battery fires is growing. In 2023, the New York City fire department responded to 268 residential fires started by batteries in e-bikes, which caused 150 injuries and 18 deaths.

These fires are so dangerous because they get blisteringly hot, almost instantly. A battery can emit a jet of flame up to 1100 C (2012 F), almost as hot as a blowtorch, and it reaches that temperature in about a second. This differs dramatically from other kinds of residential fires, which typically start slower as a smolder or a small flame. With those fires, there is more time for the smoke to reach a smoke alarm and warn people of the danger.

In addition to heating up faster, a lithium-ion battery doesn't produce much smoke to detect during the first stages of failure. By the time that limited smoke activates a traditional smoke alarm, it may be too late to prevent the fire from spreading. So it would be helpful to have a way to detect these fires early.

"While watching videos of exploding batteries, I noticed something interesting," said Tam. "Right before the fire started, the safety valve in the battery broke and it made this little noise. I thought we might be able to use that." He was not the first to make this observation, but he wanted to see if he could test the idea for himself.

Time-lapse footage of one of the experiments shows the safety valve breaking about two minutes before the battery explodes. Video by: Jian Chen/Xi'an University of Science and Technology

Before a lithium-ion battery catches fire, a chemical reaction causes pressure to build up inside. The battery starts to swell. Many lithium-ion battery cells can't expand because they have hard casings. Many of these hard casings contain a safety valve designed to break and release this pressure. This breaking safety valve is the sound Tam heard in the videos. It's a distinctive click-hiss, a little like the sound of cracking open a bottle of soda.

Previous studies have observed that sound could be used in an early warning system. But there are many sounds in the world that are similar to a breaking safety valve, such as using a stapler or dropping a paper clip. A detector would not be very useful if any random noise in a room triggered it. So the researchers needed software that could reliably recognize the sound of a breaking safety valve and not detect other noises. They decided to use AI. Tam and Putorti trained a machine learning algorithm to recognize this distinct noise.

To make the algorithm work, they first needed lots of examples of the sound. Through a collaboration with a laboratory at Xi'an University of Science and Technology, they recorded audio from 38 exploding batteries. Then they tweaked the speed and pitch of those recordings to expand them into more than 1,000 unique audio samples they could use to teach the software what a breaking safety valve sounds like.

The algorithm works remarkably well. Using a microphone mounted on a camera, the researchers detected the sound of an overheating battery 94% of the time.

"I tried to confuse the algorithm using all kinds of different noises, from recordings of people walking, to closing doors, to opening Coke cans," explained Tam. "Only a few of them confused the detector."

Tam presented the results of their study at the 13th Asia-Oceania Symposium on Fire Science and Technology. He and Putorti have applied for a patent and plan to continue this line of research with more types of batteries and microphones. In their tests, they observed that the safety valve broke about two minutes before the battery failed catastrophically. They also plan to verify that warning time with more experiments on a wider range of batteries.

Once it's fully developed, this technology could be used to build a new kind of fire alarm. These alarms could be installed in homes and office buildings, or in places with lots of batteries like warehouses and electric vehicle parking garages. The advanced warning they provide could give people time to evacuate. The key to a healthy relationship with lithium-ion batteries might just be careful listening.

Public link

Please use the above public link if you want to share this noodl on another website.