Development of a generalized state-of-health estimation framework for multi-chemistry lithium-ion batteries using incremental capacity analysis and ensemble learning

Abstract

The object of this study is the State-of-Health of lithium-ion batteries with different battery chemistries under charging conditions. The problem addressed is the limited ability of existing methods to maintain prediction of accuracy across different electrochemical characteristics and partial-charging conditions, limiting their application in battery management systems. A generalized State-of-Health estimation framework was developed using Incremental Capacity Analysis and ensemble learning methods. Charging and discharging data from Lithium Nickel Cobalt Aluminum Oxide, Lithium Iron Phosphate, and Lithium Cobalt Oxide batteries were processed using Gaussian smoothing before extracting three degradation-related features: maximum incremental capacity peak, peak voltage, and peak area. Prediction models were developed using Support Vector Regression, Extreme Gradient Boosting, and Stacking Ensemble algorithms. The results showed that the maximum incremental capacity peak and peak area had the strongest relationship with State-of-Health, with a correlation coefficient of 0.98. The Stacking Ensemble achieved the best prediction performance, with Root Mean Square Error, Mean Absolute Error, and Mean Absolute Percentage Error values of 1.89%, 1.45%, and 1.70%, respectively. Cross-material validation confirmed that the selected features remained applicable across the three battery chemistries. The proposed approach integrates Incremental Capacity Analysis-derived features with ensemble learning within a cross-material validation framework. It enables reliable State-of-Health estimation under partial-charging conditions and supports battery management systems for electric vehicles and energy storage applications