State of health estimation of lithium-ion battery cell based on optical thermometry with physics-informed machine learning

Abstract

Effective thermal management and accurate state of health (SOH) estimation of lithium-ion batteries is crucial for ensuring their safety, reliability, and longevity. This study presents three innovative physics-informed machine learning-based SOH estimation techniques trained and demonstrated using experimental temperature data. Temperature distribution measurements were obtained using optical frequency domain reflectometry with optical fibers embedded in a cylindrical lithium-ion battery cell under various SOH. One of the trained model accurately predicted the SOH of a cell within 2% with only a 10-minute measurement. This technique also enables the estimation of SOH for individual cells connected in series or parallel within a battery module or pack simultaneously, thereby reducing the overall SOH estimation uncertainty without the need for disassembly. Furthermore, this not only highlights the necessity of precise thermal management in maintaining battery health but also offers a practical and efficient solution for real-time SOH monitoring in battery systems.