Physics-informed neural networks for a Lithium-ion batteries model: A case of study

The issue related to sustainability has become central in recent years. In this field, lithium-ion batteries have been widespread, especially in the automotive field, to reduce emissions. Their use, however, requires the ability to analyse the physical patterns related to the reactions that determine the charging and discharging phase of batteries in order to monitor them and verify the life cycle. For this purpose, it is necessary to identify solving methods that allow the accurate and efficient analysis of models related to lithium-ion batteries, which consist of partial differential equations of reaction-diffusion type. Thus, this work aims to analyse the behaviour of Physics-Informed Neural Networks (PINNs) in solving differential problems related to models describing battery charging and discharging, focusing on the latest developments in the literature. Therefore, in the experimental phase, test problems are analysed on which PINNs achieve an appropriate accuracy for a qualitative analysis of the models under consideration.