FURI | Spring 2025

Predicting the Health of Lithium-ion Batteries Using Machine Learning

Lithium-ion batteries are used extensively, but their performance degrades with time. Predicting this decrease accurately and identifying how a battery can have an extended lifespan is crucial for improving longevity. Traditional ways of assessing battery health like capacity fade tests and Coulomb counting can be slow and expensive, so researchers are turning to data-driven methods. This study analyzes Electrochemical Impedance Spectroscopy (EIS) data utilizing Gaussian Process Regression (GPR), a machine learning model that effectively captures non-linear trends in battery degradation. The model analyzes EIS data to predict capacity fade and assess the impact of aging factors with improved accuracy.