Risk assessment of lithium-ion battery road transportation using the data-driven Bayesian network considering battery self-heating

Abstract

As a key component in electric vehicles or electronic devices, highly flammable lithium-ion batteries have been a growing concern for transportation safety, as evidenced by a number of lithium-ion battery fires in vehicle containers. Therefore, it is important to assess the key risk factors for fire accidents during the transportation of lithium-ion batteries. This study proposes a dynamic Bayesian assessment model for the transport risk assessment of lithium-ion batteries considering battery self-heating. A simulation model is constructed to explore the self-heating law of lithium-ion batteries and quantify their self-heating risk during transportation process. Based on Bayesian networks, a lithium-ion battery transportation risk assessment model is constructed by combining historical accident data and expert knowledge. The prediction results show that transportation accidents become more severe as transportation time increases. After four hours of transportation, the risk value was 12.28 % higher than that at the beginning. The battery self-heating and the improper handling are the most likely risk factor for fatal accidents, with a probability of 92.2 % and 72.4 % respectively, and these two risk factors have a great influence on accident severity. The mutual information of the ambient temperature is nearly four times that of the critical ambient temperature, indicating the ambient temperature is the key factor influencing the battery self-heating. This work aims to provide a theoretical basis for improving battery safety and reducing hazards during transportation.