An LLM-based framework for state of health estimation of lithium-ion batteries

Accurate state of health (SOH) estimation is essential for ensuring the reliability and safety of lithium-ion batteries. However, existing methods often suffer from complex modeling requirements and limited generalizability. Addressing these challenges calls for a shift toward knowledge-driven, scalable solutions. This study proposes a novel SOH estimation framework inspired by Large Language Models (LLMs), leveraging a token-based input structure to sequentially process battery data while preserving critical temporal dependencies. By integrating engineering informatics principles, our framework enhances modular design and universality without compromising accuracy. Specifically, it employs a feature mapping network to extract high-relevance representations, a dedicated time-series modeling module to capture dynamic degradation patterns, and a prediction head for precise SOH estimation. Rigorous experimental validation on MIT and HUST datasets demonstrates the framework’s superior effectiveness, scalability, and efficiency, highlighting its potential for real-world deployment in battery health monitoring.