Data-driven identification of lithium-ion batteries: A nonlinear equivalent circuit model with diffusion dynamics

An accurate battery model is essential for battery management system (BMS) applications. However, existing models either do not describe battery physics or are too computationally intensive for practical applications. This paper presents a non-linear equivalent circuit model with diffusion dynamics (NLECM-diff) which phenomenologically describes the main electrochemical behaviours, such as ohmic, charge-transfer kinetics, and solid-phase diffusion. A multisine approach is applied to identify the elements for high frequency dynamics, as well as a distributed SoC dependent diffusion model block is optimized to account for long time dynamics. The model identification procedure is conducted on a three-electrode experimental cell, such that NLECM-diff models are developed for each electrode to then obtain the full cell voltage. Results imply that the NLECM-diff reduces the voltage root mean square error (RMSE) by 49.6% compared to a conventional ECM in the long duration discharge and has comparable accuracy to a parameterized SPMe in the NEDC driving cycle. Additionally, the variation of diffusion-related characteristics of the negative electrode under different currents is determined as the primary reason of the battery models' large low-SoC-range error. Furthermore, the diffusion process is determined as the dominant voltage loss contributor in the long duration discharge and the ohmic voltage loss is identified as the dominant dynamic under NEDC driving profile.

Automated summary

This paper proposes a nonlinear equivalent circuit model to describe the electrochemical behaviors of batteries. The model utilizes a multisine approach for element identification and optimizes the diffusion model for SoC dependence. Experimental results demonstrate the high accuracy of the model in long duration discharge and NEDC driving cycle.