A Novel Approach for Robust and Precise Lithium-Ion Battery Degradation Investigation Through Digital Twin Using Neural Network Modeling

Rechargeable lithium-ion batteries play a crucial role in the widespread use of electric vehicles, energy systems in power grids, and portable technology devices. Their cost of aging, due to reduced charging capacity from repeated cycles, drives research into extending battery life. This research presents an innovative digital twin framework for assessing the degradation of lithium-ion batteries. Utilizing a neural network, the model simulates complete battery cell discharge, accurately determining actual discharge capacity and performance degradation from online voltage measurements. Experiments with accurate battery data confirm the method's high accuracy and robustness, with a low Mean Absolute Error value of around 0.004 during training and validation. The digital twin model allows simulation of the entire discharge process and can be scaled to form battery packs, improving maintenance and control efficiency. The experimental results showed promising outcomes, indicating the potential of this approach for extending battery lifespan. This paves the way for an analysis of parallel lithium-ion battery pack discharge, along with the creation of a digital twin model for the power system in the future.