Gêmeos digitais para baterias de veículos elétricos.

Abstract

This work presents the development of a digital twin for lithium-ion batteries employed in urban electric ground vehicles, aiming to monitor variables such as state-of-charge and temperature. The proposed methodology integrates reduced-order electrochemical models, equivalent circuit models, and thermal models, complemented by state estimation algorithms capable of processing voltage, current, and temperature data, such as the Extended Kalman Filter. To validate the digital twin, a series of pseudo-synthetic simulations were conducted across different temperature ranges, achieved by combining two complementary tools: FASTSim, employed to efficiently replicate vehicle dynamics under three standardized driving cycles—a congested urban driving profile, a highway cycle with near-constant speeds on expressways, and a dynamic route characterized by sudden accelerations and high speeds—and PyBaMM, used to model the electrochemical behavior of the battery subjected to power profiles demanded by the vehicle. The results demonstrate accurate state-of-charge predictions, with mean absolute errors below 3% under ambient temperature conditions above 10◦C, as well as vehicle residual range estimations exhibiting absolute errors around 30km. Therefore, the developed digital twin, supported by a modular architecture and combining physics-based models with pseudo-synthetic data, enables improving battery management, as well as expanding the potential for integrating electric vehicles into novel services and architectures.