Li-ion cells have been widely used in battery electric vehicles (BEV), hybrid electric vehicles (HEV), plug-in hybrid electric vehicles (PHEV), and stationary energy storage units due to their high specific energy and excellent cycle life. However, the safety of Li-ion battery (LIB) packs has become a vital issue, especially under extreme or abusive conditions. Fire and explosion due to the onset of thermal runaway often lead to catastrophic damages. Thermal failures are caused by drastic and successive processes and reactions that feed into one another, eventually leading to fire and explosion. Thus, one of the most crucial current research goals is to steer LIBs’ technology toward long-lasting and safe operation.

In this work, we employ data-driven methods to predict and potentially prevent the thermal runaway (TRA) in a LIB battery pack with various cooling systems. Specifically, we use several deep learning (DL) techniques to learn and predict the likelihood of the TRA, using thermal images acquired from the multi-physics modeling of LIBs. The multi-physics modeling approach comprises a coupled thermal, electrochemical (P2D) model¹ and degradation due to the solid electrolyte interface (SEI) formation/decomposition sub-models.² The geometrical and material properties of the battery pack are taken from the 2170 cell as presently used in some BEV. The presented multi-physics model has been formulated as a three-dimensional model to facilitate efficient and accurate numerical simulations in different conditions. Under general working conditions, the cells in a pack are exposed to a continuous charge/discharge load, where heat is generated inside the cells due to several electrochemical and degradation processes. The multi-physics modeling results in the form of the temperature profiles of the battery surface as well as the SEI formation/decomposition rate at various charge/discharge cycles are validated against the prior literature’s experimental data. Thermal images of the battery pack from different views are collected as sequences of video frames during the battery operation from multiple simulations to be used in DL.

Based upon the multi-physics modeling results, Convolutional Neural Networks (CNNs) are proposed to predict the TRA-like events.^3–5 Hyperparameters optimization has been performed to identify variables for the best performing ML method. The proposed combined multi-physics and machine learning modeling methodology provide interesting insight and predictive capabilities of the TRA. Thus, the developed combined multi-physics and ML modeling approach establishes a basis for 'on-the-fly' prediction of the TRA as well as a framework for extending machine learning methodologies to broad applications in battery failure prediction.

References

1. J. Newman and W. Tiedemann, J. Electrochem. Soc., 140, 1–5 (1993).
2. X. Fen, X. He, M. Ouyang, L. Wang, L. Lu, D. Ren and S. Santhanagopalan, J. Electrochem. Soc., 165, A3748–A3765 (2018).
3. J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recognit., 2016-Decem, 779–788 (2016).
4. K. He, X. Zhang, S. Ren, and J. Sun, Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recognit., 2016-Decem, 770–778 (2016).
5. K. Simonyan and A. Zisserman, 3rd Int. Conf. Learn. Represent. ICLR 2015 - Conf. Track Proc., 1–14 (2015).