Data-Driven State of Health Estimation of Lithium-ion Battery Cells Using Machine Learning

Many battery cells are being degraded once it is used for a period of time. Most battery cells can only show the State of Charge (SoC) of the battery when it is being charged. However, the battery cells do not have the same capacity as when they are new. Thus, knowing the State of Health (SoH) of a battery is a must. There are many machine learning methods that can be used to estimate the SoH of the battery, such as Random Forest, Gradient Boosting, Support Vector Regression, and Neural Network. These methods are being compared in this study based on the mean absolute error (MAE) of the methods. This study uses data-driven to estimate the SoH using four different machine learning methods. This study shows 0.05 for the MAE for battery cell Li-on 18650 and 0.02 MAE for battery cell Li-ion 21700. Both MAE for Li-ion 18650 and Li-ion 21700 uses RF method as the best prediction model.