AI Listens for Danger: New Tech Detects E-Bike Battery Fires Before They Ignite
Amsterdam, Friday, 15 November 2024.
Researchers have developed AI technology that can recognize the specific sound lithium-ion batteries make just before catching fire. This breakthrough could provide crucial early warnings for e-bike users, potentially preventing accidents and saving lives in urban areas where battery fires have become a growing concern.
Unveiling the Technology
The National Institute of Standards and Technology (NIST), based in the United States, has been at the forefront of this innovation. Led by researchers Wai Cheong ‘Andy’ Tam and Anthony Putorti, NIST developed an AI-powered method that detects the unique ‘click-hiss’ sound emitted by lithium-ion batteries before they catch fire. This sound is produced when gas escapes through a safety valve, a precursor to catastrophic battery failure[1][2].
How the AI System Works
The AI technology was trained using more than 1,000 unique sound samples from recorded battery explosions, a project that involved collaboration with Xi’an University of Science and Technology. This extensive dataset enabled the AI algorithm to achieve a 94% accuracy rate in recognizing the ‘click-hiss’ sound even in noisy environments typical of battery-dense areas like warehouses and garages. The system can detect failing battery sounds approximately two minutes before a potential fire, allowing for timely evacuation or intervention[2].
Applications and Future Prospects
Currently, the system is being tested for deployment in homes, offices, and electric vehicle garages, where it could serve as an early warning system. The potential to integrate this technology into specialized fire alarms presents a significant advancement in fire safety protocols. Future expansions could see this AI system applied to other battery types, broadening its applicability and enhancing safety across various sectors[2].
Addressing the Growing Battery Safety Challenge
With the rise in battery-operated devices, particularly in urban settings, the risk of lithium-ion battery fires has become a pressing issue. For instance, New York City alone witnessed 268 e-bike battery fires in 2023, resulting in 150 injuries and 18 fatalities. Traditional fire detection methods are often too slow, as battery-induced flames can reach temperatures of up to 1,100 °C. The AI technology developed by NIST offers a proactive solution by providing early warnings, potentially preventing such incidents[1][2].