Safety of Li ion batteries is as important, if not more, as its performance because its usage in contemporary transport applications can otherwise lead to catastrophic scenarios. Owing to the exothermic reactions and complex heat transfer within the Li ion batteries, they are prone to overheating and possible thermal runaway in the absence of adequate thermal management systems. Therefore, to ensure a safe operation it is key to evaluate the uncertainties with respect to a thermal management system along with the uncertainties in the intricate multiphysical phenomena within the system. To achieve this, we conduct Monte Carlo simulations followed by sensitivity analysis to correlate the variability of 14 different factors with the safety of an 18650 Sony cell with a phase change material (PCM) at stressed conditions. These varied factors correspond to the properties of the 18650 battery and the PCM. From the PCM properties, we identify the temperature at which the PCM starts melting to have most influence on the safety of the battery. In addition, we also estimate the probability of the sub-optimal unsafe scenarios occurring by examining the distribution of the maximum temperature within the battery. Finally, we obtain reduced surrogate models with an accuracy of 92% through supervised machine learning algorithms.

锂离子电池的安全性甚至与其性能一样重要，因为它在现代运输应用中的使用可能会导致灾难性的后果。由于锂离子电池内部的放热反应和复杂的热传递，在缺乏适当的热管理系统的情况下，它们易于过热并可能导致热失控。因此，为了确保安全运行，评估与热管理系统有关的不确定性以及系统中复杂的多物理现象的不确定性至关重要。为了实现这一目标，我们进行了蒙特卡洛模拟，然后进行灵敏度分析，以将14种不同因素的变异性与18650索尼电池在相变材料在压力条件下的安全性联系起来。这些不同的因素对应于18650电池和PCM的属性。根据PCM的属性，我们确定PCM开始熔化的温度对电池的安全性影响最大。此外，我们还通过检查电池内部最高温度的分布来估计发生次优不安全情况的可能性。最后，我们通过监督的机器学习算法获得精简的替代模型，其准确度达到92％。我们还通过检查电池内部最高温度的分布来估计发生次优不安全情况的可能性。最后，我们通过监督的机器学习算法获得精简的替代模型，其准确度达到92％。我们还通过检查电池内部最高温度的分布来估计发生次优不安全情况的可能性。最后，我们通过监督的机器学习算法获得精简的替代模型，其准确度达到92％。