Predicting Lithium-Ion Battery Degradation Modes Using a Machine Learning Approach Based on EIS Measurements

This study presents a machine learning (ML) framework that is able to predict the degradation modes of a lithium-ion battery. Leveraging on some analytical models recently proposed in literature, the proposed algorithm estimates three main ageing indicators (Conductivity Loss (CL), Loss of Active Material (LAM) and Loss of Lithium Inventory (LLI)), using ML models based on battery-related features, i.e., Electrochemical Impedance Spectroscopy (EIS), State of Charge (SoC), Open Circuit Voltage (OCV) and number of cycles. In order to test the accuracy of these ML models, two different datasets were created (for testing and validation), using data acquired from 8 different cells. Experimental results show that the proposed approach, based on ML, is able to determine the ageing status of a battery with high accuracy, thus offering a reliable solution for battery health diagnostics.