Machine Learning-Based Framework for Online State of Health Assessment and End-of-Life Prediction in Commercial Lithium-Ion Batteries

Lithium-ion batteries (LIBs) are an essential and versatile energy source for portable electronics, grid energy storage, and electric vehicles. To meet the increasing demand for energy, it is crucial to understand physical changes to further improve LIBs performance and safety. State of health (SOH) is a metric that indicates how safe LIBs are for operation. This work presents a comprehensive framework for SOH assessment of LIBs. The proposed framework was validated using data from commercial LIBs tested at 1C/1C (charge/discharge), at different temperatures and different depths of discharge (DOD). Reference performance tests (RPTs) were periodically conducted until lithium-ion cells (LICs) reached the end of life. Differential voltage analysis (DVA) was used to acquire relevant physical parameters from RPTs. The continuous monitoring of DVA provided updates about LIC degradation based on internal resistance, slippages, active masses, and stoichiometries of both electrodes during the cycling process. For prediction of degradation mechanisms and end of life (EOL), a machine learning model based on Random Forest was integrated into a Raspberry Pi computer. This affordable computer enabled remote monitoring of LICs performance, which transmitted SOH and degradation data to a cloud-based service. Results indicate that LICs cycled at 100% DOD experienced greater cathode material loss compared to those cycled at lower DODs. The RF model, running on a Raspberry Pi computer, achieved a processing time of approximately 2.9 seconds and demonstrated an accuracy of 86.67%. This attained accuracy indicates the potential of this affordable framework to be employed in real-time (online) assessments of SOH in either small or larger LIBs with different chemistries.