Thermal performance prediction of the battery surface via dynamic mode decomposition.

The heat dissipation from the battery surface significantly affects battery performance and lifetime. This study proposes a new and an alternative method to predict the thermal performance of the battery operation according to the surface temperature gradients and heat & exergy losses by using a data-driven dynamic mode decomposition method, which is new for thermal flows. To predict the thermal gradients, a 10 min long experiment is performed via an infrared thermographic camera for a commercial Li-polymer battery of a smartphone. The camera collects the thermal images on the battery surface along 1 min as the data training period at first; then, the proposed method predicts the surface temperature gradients for the rest of the experimental period, 5 min. The temperature gradients on the battery surface are well predicted with less than 1% error whereas the heat dissipation and the exergy loss are predicted with the maximum error values of 2.75% and 5.30%, respectively. According to the error probability distribution plots, the vast majority of the occurred error is less than ±5%. The results prove the fast prediction ability of the proposed technique and show promising outcomes for further improvement studies. Image 1 • A novel data-driven technique is developed for heat loss from batteries. • Training periods of 30 and 60 s-periods achieve well-prediction. • The method is verified via experimental results with less than 5% errors. • Data visualization is shown for 5 min-long battery charging. [ABSTRACT FROM AUTHOR]