Capacity Estimation and Knee Point Prediction Using Electrochemical Impedance Spectroscopy for Lithium Metal Battery...

Area of Science:

• Materials Science
• Electrochemistry
• Data Science

Background:

• Lithium-metal batteries (LMBs) offer high energy density but suffer from complex degradation, hindering performance prediction.
• Accurate health monitoring is crucial for reliable operation and advanced battery management systems.

Purpose of the Study:

• To develop a machine learning (ML) framework for predicting capacity degradation and detecting the knee point (KP) in LMBs.
• To utilize electrochemical impedance spectroscopy (EIS) data combined with ML for enhanced battery prognostics.

Main Methods:

• A machine learning framework integrating EIS data with the XGBoost algorithm was developed.
• SHapley Additive exPlanations (SHAP) analysis identified key impedance features influencing degradation prediction.
• Input data was optimized by selecting critical frequency points based on SHAP values.

Main Results:

• The ML models accurately estimated capacity degradation and detected the KP in LMBs.
• Low-frequency impedance components were crucial for capacity estimation, linked to diffusion processes.
• High-frequency impedance features dominated KP detection, relating to charge transfer resistance.
• Optimized models showed comparable or improved accuracy with reduced data complexity.

Conclusions:

• EIS-based ML models provide accurate forecasting of LMB health and aging.
• SHAP analysis offers valuable insights into degradation mechanisms.
• The framework enhances battery management strategies through improved prognostics.