A control-oriented electro-thermal-mechanical modeling method for lithium-ion batteries considering aging effects

The modeling and application of lithium-ion batteries (LiBs) multi-physics characteristics can significantly enhance battery health assessment and thermal runaway early warning capabilities. However, existing multi-physics modeling studies that involve the mechanical properties of batteries primarily focus on either the modeling of internal electrode deformation mechanisms or external mechanical characteristics—rarely both in an integrated manner. There is a notable lack of comprehensive approaches that simultaneously consider internal electrode strain behavior and external mechanical responses. Such integrated modeling is especially valuable for application scenarios where external mechanical characteristics are used to infer internal strain behavior. Moreover, limited efforts have been made to simplify models in a control-oriented manner tailored to the piecewise monotonic mechanical behavior of LiFePO4 (LFP) batteries. In this study, an electro-thermal-mechanical model (ETMM) centered on a piecewise nonlinear volumetric strain model is proposed to accurately capture the dynamic mechanical behavior of LFP batteries. Within the proposed framework, the electro-thermal coupling model provides the state of charge (SOC) and battery temperature as inputs to the mechanical model, which then outputs the mechanical expansion force via an equivalent force estimator. Specifically, the model captures the propagation of volumetric changes induced by lithium-ion insertion and extraction within the electrode materials, ultimately manifesting as observable external mechanical responses. Furthermore, the influence of battery aging on model parameters is considered, and a hybrid offline-online parameter identification strategy is employed for parameter calibration. Finally, the proposed model is validated under various ambient temperatures, aging conditions, and dynamic operating scenarios. Under the specified conditions, experimental results show that the proposed model achieves an MAE of 5.38 N and an RMSE of 6.52 N. Compared to related work, this reflects a 2.01 N improvement in MAE, highlighting enhanced predictive precision and modeling reliability. Additionally, the model’s low computational cost highlights its potential for real-world engineering applications.