Investigation of Immersion Cooling Efficiency for Lithium-Ion Battery Pack with Machine Learning Validation
Battery thermal management systems play a critical role in improving their power capability, extending lifetime, and minimizing the probability of thermal runaway. These systems aim to maintain a regulated operating temperature by removing heat from the battery pack and homogenizing the temperature within and between the cells. Precisely predicting the outlet fluid temperature is crucial to enhance the efficiency and functionality of these systems through proper design and operation. In this study, we applied machine learning techniques to predict the outlet fluid temperature of immersion cooling for a lithium-ion battery pack. Nine regression models have been employed to accurately predict the outlet fluid temperature. The results indicate that the Light GBM model can significantly enhance the efficiency and performance of immersive cooling systems for Li-ion batteries. We validated our approach by comparing the predicted temperature distributions with experimental data with MAE and RMSE of 0.0136 and 0.0209, which demonstrated that our model effectively captures the heat transfer in the battery pack.
Paper
Full text
Investigation of Immersion Cooling Efficiency for Lithium-Ion Battery Pack with Machine Learning Validation
Semantic Scholar · Engineering · 2023
Abstract
Battery thermal management systems play a critical role in improving their power capability, extending lifetime, and minimizing the probability of thermal runaway. These systems aim to maintain a regulated operating temperature by removing heat from the battery pack and homogenizing the temperature within and between the cells. Precisely predicting the outlet fluid temperature is crucial to enhance the efficiency and functionality of these systems through proper design and operation. In this study, we applied machine learning techniques to predict the outlet fluid temperature of immersion cooling for a lithium-ion battery pack. Nine regression models have been employed to accurately predict the outlet fluid temperature. The results indicate that the Light GBM model can significantly enhance the efficiency and performance of immersive cooling systems for Li-ion batteries. We validated our approach by comparing the predicted temperature distributions with experimental data with MAE and RMSE of 0.0136 and 0.0209, which demonstrated that our model effectively captures the heat transfer in the battery pack.