Inlet setting strategy via machine learning algorithm for thermal management of container-type battery energy-storage systems (BESS)

Abstract

This research enhances the safety and efficiency of the container-type battery energy storage systems (BESS) through the utilization of machine learning algorithms. The decision tree algorithm and support vector machine (SVM) are employed to clarify the influence of cooling air on temperature distribution and predict the safety of battery modules. The results indicate that the flow rate of cooling air has significant impact on both the maximum temperature and temperature difference of the batteries, while the inlet temperature only affects the maximum temperature. Additionally, the distributed inlet setting strategy is introduced for the container-type BESS. The strategy effectively reduces the temperature of the target battery by 8.6 % and decreases the highest temperature in the BESS by 3.2 % without creating new hotspots. Comparing with the uniform inlet setting strategy, this strategy reduces the power consumption by 40 %, providing great efficiency for cost consideration. In conclusion, this research clarifies the relationship between the conditions of cooling air and the temperature distribution of batteries using the machine learning algorithms, and introduces the distributed inlet setting strategy to prevent thermal runaway in the BESS with less power consumption.