Sensitivity analysis methodology for battery degradation models

Accurate degradation models are crucial to perform efficient battery design and management. The time resources required improve the output accuracy of depends critically on ability assess sensitivity input factors governing inherent dominant mechanism in model. Here, we present a analysis pseudo-two-dimensional model coupled with capacity fade based solid-electrolyte interphase (SEI) formation corresponding irreversible charge loss for Li-ion batteries. proposed method is training an inexpensive differentiable surrogate Gaussian process regression observed input–output pairs analysing learn about global local sensitivities original system. With this method, relevant sensitive parameters can be identified, in-depth electrochemical phenomena such as correlation between thickness SEI explored. will provide key insight into how physics-based must conducted effective integration management systems.