AI Could Help Detect Lithium-Ion Battery Fires Before They Ignite

Image by Pixabay

Lithium-ion batteries power many of today’s devices, from smartphones to electric vehicles. However, they also pose a significant fire risk, particularly when they overheat or fail. To address this threat, researchers have developed a groundbreaking method to detect when these batteries are about to catch fire—by listening for a specific sound.

The concept stems from the fact that, before a lithium-ion battery catches fire, a chemical reaction inside the battery causes pressure to build up. This pressure causes the battery to swell, and when the casing can’t expand, a safety valve is triggered. This valve breaks, releasing the pressure and making a distinctive “click-hiss” sound. Recognizing this sound could provide a critical early warning of an impending fire.

Researchers at the National Institute of Standards and Technology (NIST), in collaboration with a lab from Xi’an University of Science and Technology, have trained a machine learning (ML) algorithm to recognize this sound. They fed the algorithm over 1,000 audio samples, including recordings from 38 exploding batteries, and used them to teach the software how to identify the specific sound of a breaking safety valve.

The results have been promising. The algorithm successfully detected the sound of an overheating battery 94% of the time using a microphone mounted on a camera. The researchers even tested it with various background noises—like people walking, doors closing, and soda cans opening—and found that the algorithm still performed well, with only a few false positives.

This technology could pave the way for new fire detection systems that provide early warnings of thermal runaway in lithium-ion batteries. Such systems could be installed in homes, offices, warehouses, and electric vehicle parking garages, offering extra time to evacuate or address potential fire hazards before they escalate.

The ultimate goal is to create a fire alarm system that listens for these early warning sounds, providing a potentially life-saving layer of protection for environments where large numbers of lithium-ion batteries are used.

With this technology, the future of battery safety might not just be in better designs but also in careful listening to the sounds of potential danger.