Scientists at the National Institute of Standards and Technology (NIST) have developed a novel approach to detect impending lithium-ion battery fires by analyzing sound. This innovative technology leverages artificial intelligence (AI) to recognize the distinct noise made by a breaking battery safety valve—providing crucial warning time before a fire occurs.
Why Early Detection Matters
Lithium-ion batteries are everywhere, powering phones, laptops, e-bikes, and electric cars. They are compact and energy-dense, making them indispensable for modern life. However, these qualities also make them prone to catastrophic failures, such as fires or explosions, when damaged or overheated.
In 2023 alone, the New York City Fire Department reported 268 residential fires caused by e-bike batteries, leading to 150 injuries and 18 fatalities. These fires are particularly dangerous due to their rapid onset and extreme heat, reaching temperatures of up to 1100°C (2012°F) in just seconds. Traditional smoke alarms often fail to provide adequate warning for lithium-ion battery fires, which emit little smoke during their initial failure stages.
Observing the Warning Signs
NIST researcher Wai Cheong “Andy” Tam identified a potential early warning signal while watching videos of exploding batteries. Just before a fire starts, the battery’s safety valve—a feature designed to release pressure—emits a distinct “click-hiss” sound. Tam and his colleague, Anthony Putorti, sought to test whether this sound could reliably signal impending battery failure.
“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.
Leveraging AI for Sound Detection
To develop their detection system, Tam and Putorti collaborated with Xi’an University of Science and Technology, recording audio from 38 exploding batteries. They augmented these recordings, altering pitch and speed to create over 1,000 unique samples for training a machine-learning algorithm.
The AI-based system demonstrated impressive accuracy, identifying the sound of an overheating battery 94% of the time using a microphone mounted on a camera.
“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.”
Two-Minute Warning
In their experiments, the safety valve’s distinct sound occurred approximately two minutes before catastrophic battery failure, offering valuable evacuation time. “While the system isn’t perfect yet, our goal is to extend that warning time further and integrate this technology into practical fire alarms,” said Tam.
Applications and Future Development
Once fully developed, the technology could revolutionize fire safety systems in homes, offices, warehouses, and electric vehicle parking facilities. The advanced warning it provides could save lives by allowing occupants to evacuate before a fire spreads.
Tam presented the team’s findings at the 13th Asia-Oceania Symposium on Fire Science and Technology, and they have filed a patent application. Future research will focus on testing the system with various battery types and microphones and verifying warning times through additional experiments.
“The next step is to refine the technology and explore its potential applications in environments with high concentrations of lithium-ion batteries,” Putorti added.
As this sound-based detection technology advances, it promises to redefine the relationship between safety and lithium-ion battery usage—proving that listening carefully could save lives.
Source/Photo Credit: National Institute of Standards and Technology (NIST)
(Editor’s Note: All trademarks mentioned in this article, including company names, product names, and logos, are the property of their respective owners. Use of these trademarks is for informational purposes only and does not imply any endorsement.)
