Skip to content

Data-driven prediction of battery cycle life before capacity degradation

Nature Energy

Accurately predicting the lifetime of complex, nonlinear systems such as lithium-ion batteries is critical for accelerating technology  development.  However,  diverse  aging  mechanisms,  significant  device  variability  and  dynamic  operating  conditions  have remained major challenges. We generate a comprehensive dataset consisting of 124 commercial lithium iron phosphate/graphite  cells  cycled  under  fast-charging  conditions,  with  widely  varying  cycle  lives  ranging  from  150  to  2,300  cycles.  Using  discharge voltage curves from early cycles yet to exhibit capacity degradation, we apply machine-learning tools to both predict and classify cells by cycle life. Our best models achieve 9.1% test error for quantitatively predicting cycle life using the first 100 cycles (exhibiting a median increase of 0.2% from initial capacity) and 4.9% test error using the first 5 cycles for classifying cycle life into two groups. This work highlights the promise of combining deliberate data generation with data-driven modelling to predict the behaviour of complex dynamical systems.