Page 102 - EER-2-1

Explora: Environment
and Resource Artificial neural networks

Table 5, with an average error of approximately 2%. This unknown inputs and long-term cycling data. 28,55,57 In
confirms the reliability of the thermal model, which uses this study, both static and dynamic profiles were used to
only Joule heating to predict temperature changes. validate our lifetime model.
In summary, the electrothermal model accurately 3.4.1. Static profile validation
captures the electrical and thermal performance of the
Sony 3 Ah cell under dynamic load conditions, making it For the static profile validation, an independent dataset not
a reliable tool for optimization algorithm development. included in the ANN training process was employed. These
data were obtained from condition II (25°C; 1C discharge/0.5C
3.4. Lifetime model charge, 100% depth of discharge [DoD]; Table 2).
The validation of an aging model typically involves Figure 8A presents a comparison between the measured
comparing its lifetime predictions against a set of previously and simulated capacity retention over 1400 FECs. An

A B

C D

E F

Figure 10. Validation of the coupled electrothermal lifetime model for condition II: (A) voltage comparison of 500 FECs; (B) temperature comparison
of 500 cycles; (C) voltage comparison of 1000 FECs; (D) temperature comparison of 1000 cycles; (E) voltage comparison of 1500 FECs; (F) temperature
comparison of 1500 FECs.
Abbreviation: FECs: Full equivalent cycles.

Volume 2 Issue 1 (2025) 9 doi: 10.36922/eer.7228