Battery cumulative lifetime prognostics to bridge laboratory and real-life scenarios

Abstract

Battery operating data from real-life scenarios are riddled with randomness, complexity, and multi-cell grouping, posing significant challenges for applying lifetime prognostic approaches developed from laboratory scenarios. To address this, we have conducted extensive experimental investigations into battery degradation across laboratory and real-life scenarios spanning a 4 year period, involving a total of approximately 546,000 charge-discharge cycles across hundreds of cells and packs. In addition to our experimental investigations, we develop a lifetime prognosis approach by creatively incorporating the concept of cumulative utilization lifetime. Our approach highlights the significant potential of transferring knowledge gained from standardized laboratory scenarios to diverse real-world conditions. It consistently improves performance from early prediction to real-time prediction, achieving a remarkable error margin of around 5% and millisecond-level computational efficiency on a portable laptop with no dedicated graphics. Furthermore, our experimental investigations underscore the beneficial effects of seasonal low temperatures on prolonging battery lifetime.