基于AI技术的锂离子电池存储区火灾早期探测技术研究

Research on early fire detection technology of lithium-ion battery storage area based on AI technology

为快速探测锂离子电池存储区域内失效的电池,分析存储区内锂离子电池的失效模式及火灾表征因子,利用基于图像识别和大数据Artificial Intelligence(AI)的技术,对锂钴电池、三元电池、三元电池组(PACK)的堆垛进行理论分析及模拟测试。结果表明:锂离子电池堆垛失效可分为6个阶段:锂离子电池失效外表面温度缓慢上升阶段、乳白色气体沿地平面飘浮阶段、黑色烟气上升阶段、箱体外温度达到探测阈值阶段、着火起始及传播阶段和燃尽阶段;提出了锂离子电池堆垛失效火灾表征因子出现的顺序为白雾、烟气、温度及火焰;同时,开发了一种适用于锂离子电池存储区白雾、烟气及火焰的基于图像识别与大数据分析的AI探测系统,且该系统可在冒白雾1 min内有效预警,较吸顶感烟火灾探测器响应时间快5~10 min。

In order to quickly detect the failed batteries in the storage area of lithium-ion batteries, the failure modes and fire characterization factors of lithium-ion batteries in the storage area are analyzed. Failure detection test of lithium cobalt batteries, ternary batteries and ternary battery packs were carried out by using the technology of artistic intelligence(AI) based on image recognition and big data. The results show that the stacking failure of lithium-ion batteries can be divided into six stages:the sharp rise of lithium-ion battery temperature, the ripple of milky white gas along the ground plane, the rise of black smoke, the initiation of ignition, flame expansion and burnout.The fire characterization factors of lithium-ion battery stacking failure are white fog, smoke, temperature and flame. At the same time, an AI detection system based on image recognition and big data analysis suitable for lithium-ion battery storage area fire is developed, and the system can realize early warning within 1 min of white fog, and the response speed is 5 ~ 10 min faster than that of ceiling smoke fire detector.