KI kann „hören“, wenn eine Lithium-Batterie im Begriff ist, Feuer zu fangen: „Wenn eine Lithium-Ionen-Batterie im Begriff ist, Feuer zu fangen, gibt sie ein einzigartiges Klick-Zischen von sich, wenn Gase entweichen.“ NIST-Forscher haben KI darauf trainiert, dieses Geräusch auch in lauten Umgebungen zu erkennen. »
https://www.nist.gov/news-events/news/2024/11/ai-can-hear-when-lithium-battery-about-catch-fire
7 Comments
« 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. »
Reference: Tam, W. , Chen, J. , Tang, W. , Tong, Q. , FANG, H. and Putorti Jr., A. (2024), Development of a Robust Early-Stage Thermal Runaway Detection Model for Lithium-ion Batteries, 13th Asia-Oceania Symposium on Fire Science and Technology, Daegu, KR, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=958183
Couldn’t you do the same thing with an audio filter ?
“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.”
“right before” needs to be quantified. not even the linked study has any mention of time. so what good is it for detection if it’s 0.3 seconds. that’s not enough time to react regardless if AI can detect or not.
“Alert: your phone is about to become a grenade. Please toss it towards someone you really don’t like”
but if the valve burst 2 minutes prior to the battery iginiting, doesn’t that mean the battery’s internal pressure collapsed? or that pressure had accumulated which in turn led to it bursting?
Having a microphone mounted on a camera to detect failures sounds doable, but in a multipack battery like in EVs what do you do? a microphone and camera for each cell? or per module? or per pack?
I can’t quite see how this would go into production. not to mention the way they tried to jam the algorithm that doesn’t cover real life usages.
More research indeed would be useful, but the equipment, just adds to the car price.
Wouldn’t it be easier/cheaper to put a sensor inside the battery that can detect the gases as they accumulate there *before* the safety valve blows.