Lithium-ion battery aging: An integrated review of mechanisms, multi-scale modelling, and physics-informed health prediction

초록

Lithium-ion battery (LIB) aging is a complex multi-physics phenomenon that poses challenges for performance, longevity, and safe operation. This work presents a holistic review connecting fundamental aging mechanisms, multi-scale modeling, and state-of-the-art health prediction. First, intrinsic and extrinsic aging factors are elucidated with emphasis on critical operational thresholds governing dominant mechanism transitions. For instance, the same LFP/graphite cell shows a 45 °C threshold for its dominant mechanism shift from lithium loss to cathode degradation at a fixed C/3 rate; yet a 4C threshold at a fixed 25 °C. Such thresholds are not universal constants but vary with the operating conditions and cell design. Building upon these mechanism-consistent boundaries, design strategies for the accelerated aging tests are proposed to ensure physically representative datasets. Second, multi-scale modelling strategies are evaluated across atomistic, electrochemical, and system-level scales. The embedding of heterogeneous SEI energy barriers from molecular dynamics and DFT into electrochemical models is highlighted as key to enhancing parameterization accuracy, while the coupling among battery, thermal, and aging models is illustrated to improve framework fidelity. Finally, methods for predicting State of Health (SOH) and Remaining Useful Life (RUL) are quantitatively benchmarked across model-based, data-driven, and physics-informed hybrid approaches. Physics-informed neural networks lower RUL prediction error (MARE from 5.71% to 3.19% using only early-cycle features) and reduce data requirements under data-scarce conditions, while discovery learning reduces validation time by up to 98%. Cloud-edge collaborative frameworks are identified as the most promising architecture for practical BMS deployment. Selection guidelines are provided for developing next-generation battery management systems.

키워드