Thermo-mechanical behavior evolution analysis and fusion SOC estimation of cylindrical LiFePO4 batteries

Hits:

Impact Factor:9.4

DOI number:10.1016/j.energy.2025.138867

Journal:Energy

Key Words:Cylindrical LiFePO4 batteries; Thermo-mechanical behavior evolution; Fusion SOC estimation; Interpretability analysis

Abstract:The intricate coupling among electrical, thermal, and mechanical dynamics in cylindrical LiFePO4 batteries poses significant challenges to accurate SOC estimation. To address this, our study conducts a comprehensive comparative analysis of the thermal-mechanical behavior evolution characteristics of cylindrical LiFePO4 batteries, across a wide temperature range and different aging stages. The core innovation of this work lies in the development of a fusion SOC estimation methodology that integrates electro-thermal-mechanical behaviors. A novel contribution rate-behavior mapping function is introduced to quantitatively evaluate and visualize the influence of each physical behavior on SOC estimation, thereby significantly enhancing its transparency and interpretability. Experimental observations reveal that both thermal and strain dynamics exhibit distinct patterns, and strain demonstrates strong SOC-dependent characteristics, with evident asymmetry between charge and discharge processes. Furthermore, the aged battery demonstrates more pronounced fluctuations in temperature and strain, with mechanical behavior showing higher sensitivity to ambient temperature variations compared to the fresh battery. The proposed fusion method achieves remarkable SOC estimation accuracy emulating real-world scenarios: at 0.1 Hz, the mean absolute error (MAE) and root mean square error (RMSE) remain below 1.10 % and 2.00 %, at 0.03 Hz, MAE and RMSE are maintained below 1.70 % and 2.30 %. Compared with no-mechanical behavior fusion, it achieves average reductions of 72.68 % in MAE and 69.69 % in RMSE at 0.1Hz, and the improvements reach 65.11 % and 63.96 % at 0.03Hz. Comparative with conventional data-driven methods, it achieves average reductions of 30.10 % in MAE and 31.51 % in RMSE at 0.1Hz, and the improvements reach 28.82 % and 29.62 % at 0.03Hz.

Note:中科院1区Top

Co-author:Changpeng Tan

First Author:Muyao Wu

Indexed by:Journal paper

Correspondence Author:Li Wang

Document Code:138867

Discipline:Engineering

Document Type:J

Volume:338

ISSN No.:0360-5442

Translation or Not:no

Date of Publication:2025-10-15

Included Journals:SCI、EI

Links to published journals:https://www.sciencedirect.com/science/article/pii/S0360544225045098?ref=pdf_download&fr=RR-2&rr=98eb413a9d0948b7