关键词: 插电混动汽车；锂离子电池；电池健康状态；机器学习；深度学习

Abstract: State of health (SOH) serves as a key indicator for assessing the overall condition and remaining capacity of plug-in hybrid electric vehicle (PHEV) batteries. Accurate SOH prediction is of significant value for ensuring safe and reliable vehicle operation, optimizing performance, and extending battery lifespan. This paper systematically reviews the current research status of SOH prediction techniques for PHEVs, categorizing the technical approaches into experimental methods, model-based methods, data-driven methods, and model fusion methods, while summarizing the advantages and limitations of each approach based on current research. Furthermore, the paper synthesizes industry application trends in SOH prediction and management adopted by automotive manufacturers, and identifies challenges in data acquisition and processing, model accuracy and generalization, real-time performance and computational efficiency, as well as personalized modeling and user behavior integration. By elucidating the theoretical framework and practical pathways of SOH prediction techniques for PHEV batteries, this study provides critical theoretical guidance for the industry, facilitating the optimization of battery management systems, enhancement of vehicle safety and reliability, and promotion of high-quality development in the new energy vehicle market.

Key words: plug-in hybrid electric vehicles; lithium-ion batteries; state of health; machine learning; deep learning