The safety risks of lithium battery fires are often concerning. To address this, scientists have proposed a method that uses sound to provide early warnings of battery fires. Research has found that lithium-ion batteries undergo a series of chemical reactions before catching fire, which gradually increases the internal pressure of the battery, ultimately leading to battery swelling.
Image Source Note: Image generated by AI, image authorized by service provider Midjourney
The battery casing is usually rigid and cannot accommodate this swelling. Therefore, the safety valve inside the battery will rupture when the pressure becomes too high, producing a distinctive sound. This sound is somewhat similar to the pop and hiss of opening a soda bottle.
In response, a research team from the National Institute of Standards and Technology (NIST) in the United States developed a machine learning algorithm specifically designed to identify this unique rupture sound. During the algorithm training process, researchers collaborated with a laboratory at Xi'an University of Science and Technology to collect audio data from 38 exploding batteries. By adjusting the speed and pitch of these audio recordings, the research team generated over 1,000 unique audio samples to further train the algorithm.
Test results indicate that this algorithm can identify the rupture sound of overheated batteries with an accuracy of 94%. Notably, the researchers also introduced various background noises during testing, including footsteps, door slams, and bottle openings, and found that only a few noises interfered with the algorithm's judgment. This finding demonstrates the robustness of the algorithm.
The research team stated that this technology has the potential to be applied in the development of a new type of fire alarm that can be installed in various locations such as homes, offices, warehouses, and electric vehicle parking lots. By providing early alerts, this technology can give people ample time to evacuate, ensuring personal safety.
Key Points:
🔋 The research team utilizes sound recognition technology to provide early warnings of lithium battery fires, ensuring safety.
🎧 The machine learning algorithm achieves a testing accuracy of up to 94%, demonstrating good robustness.
🚨 There is potential to develop a new type of fire alarm for widespread use in various locations, providing safety assurance for people.