A Step-by-Step Time-Discrete Pinn for a Lithium-Ion Battery Model
Please login to view abstract download link
Over the past few years, battery development and optimization research has progressed significantly. For example, it is important to study the State Of Health (SOH) and State Of Charge (SOC) in automotive applications. In particular, all-state-solid lithium-ion batteries represent the cutting edge of technological developments and are the basis for powering electric cars. Providing reliable numerical simulations for the related mathematical models can allow the optimization of battery use, also avoiding potentially dangerous situations. A central issue of numerical methods is to exploit data acquired through smart sensors from the Internet of Things. This work aims to propose a methodology to gain insights into the models related to all-state-solid lithium-ion batteries through physical knowledge and supporting data. For this purpose, in this work we consider Time-Discrete Physics-Informed Neural Networks which exploit a step-by-step solving strategy that integrates sensor data. Numerical experiments testify the good accuracy and efficiency of the proposed methods.
