文献类型：期刊文献

英文题名：Towards practical data-driven battery state of health estimation: Advancements and insights targeting real-world data

作者：Chen, Hongxu[1];Chen, Ying[1,2];Sun, Changzheng[1];Huo, Liping[1,3];Luan, Weiling[1];Chen, Haofeng[1,2];Huang, Lvwei[4];Zhang, Wenjun[4];Shen, Ping[4]

机构：[1]East China Univ Sci & Technol, Sch Mech & Power Engn, Key Lab Adv Battery Syst & Safety CPCIF, Shanghai 200237, Peoples R China;[2]Inst Aircraft Mech & Control, Shanghai 200237, Peoples R China;[3]East China Univ Sci & Technol, Inst Sci & Technol Informat, Shanghai 200237, Peoples R China;[4]SAIC Motor R&D Innovat Headquarters, Shanghai 201804, Peoples R China

年份：2025

卷号：110

起止页码：657

外文期刊名：JOURNAL OF ENERGY CHEMISTRY

收录：;EI(收录号：20253118872166);WOS:【SCI-EXPANDED(收录号:WOS:001583911900001)】；

基金：This work was supported by the National Natural Science Foun-dation of China (52375144 and 52205153) , the Shanghai Pujiang Programme (23PJD019) , and the Shanghai Gaofeng Project for University Academic Program Development.

语种：英文

外文关键词：Lithium-ion battery; Electric vehicle (EV); State of health (SOH); Real-world application; Data-driven; Battery health management

摘要：Accurate state of health (SOH) estimation is a cornerstone for ensuring the safety, performance and longevity of lithium-ion batteries, especially in electric vehicle (EV) applications. While numerous studies have demonstrated the significant advantages of data-driven methods in SOH estimation, most rely on laboratory-standardized test data. This raises concerns about the generalization and robustness of the models under real-world operating conditions, where batteries undergo irregular driving patterns, incomplete charging cycles, and unpredictable environments. Notably, real-world EV data reflects the coupling between battery aging characteristics and actual operating conditions, providing an unprecedented perspective for developing SOH estimation models. This review provides a comprehensive and systematic overview of data-driven SOH estimation using real-world data, a topic that has received increasing attention but lacks a consolidated research framework. The paper begins by reviewing the established SOH estimation methodologies and points out the specific challenges arising from the transition to real-world data. It then probes practical issues across the pipeline: data pre-processing for anomalies, solutions for the lack of labels, feature extraction from complex operating data, machine learning model construction, and performance evaluation across various system deployments. Key insights are presented on how to handle noisy, unlabeled, and heterogeneous data using robust modeling strategies. Moreover, a valuable extension focusing on applying the advancements to battery reuse and recycling is discussed, with the goal of developing a whole lifecycle health diagnosis framework. The paper concludes with promising prospects, encompassing open-source standardized dataset establishment, weakly supervised learning, physics-reinforced modeling, real-world deployment, and advanced sensing technology, emphasizing that real-world data makes the transition of data-driven methods from theoretical validation to industrial deployment promising. This paper aims to assist researchers and practitioners in navigating the complexities of real-world SOH estimation, accelerating the collaborative innovation and industrial adoption in battery health management. (c) 2025 Science Press and Dalian Institute of Chemical Physics, Chinese Academy of Sciences. Published by Elsevier B.V. and Science Press. All rights are reserved, including those for text and data mining, AI training, and similar technologies.