Stochastic Modeling of Lithium-Ion Battery Aging for Predictive Maintenance and Advanced Health Management (RPA Pilot / Seed Project)

For energy storage, lithium-ion (Li-ion) batteries have been widely utilized in the commercial and industrial lives today, from smartphones and computers to electric tools and vehicles. Given that batteries’ performance significantly affects these devices’ usage experience, accurate monitoring of batteries’ operating conditions and prediction of their remaining service lives become a topic of high interest for the research community and industry, for ensuring reliable functioning of the devices and systems that batteries provide the power for. Successful execution of this research will advance our understanding of the effects of batteries’ architecture and usage patterns (i.e. charge-discharge cycles) on batteries’ service life, which potentially lead to improved battery charging and safety algorithms, more efficient battery pack designs, and predictive maintenance for extended battery life. The performance degradation of Li-ion batteries due to aging is greatly affected by the operating conditions and environmental temperature. It is non-linear and non-stationary, accompanied by sudden and steep capacity drops, and varies from case to case. These challenges cause current battery health management systems to fail to accurately track the batteries’ performance degradation and predict their remaining service life. To address these challenges, the proposed solution in this research will include two parts: 1) Lévy process-based stochastic modeling of battery aging and performance degradation, by taking various uncertainties into consideration; 2) Bayesian filtering-enabled real-time performance tracking, which can deal with both gradual degradation and transient performance drops. The findings from this research can be leveraged for generation of highly competitive proposals on advanced battery health management and optimization.