New AI framework predicts lithium-ion battery lifespan using continuous aging trajectory analysis
A new framework utilizing artificial intelligence has been created to forecast the aging patterns of lithium-ion batteries using diverse datasets. This method incorporates continuous voltage-capacity and capacity-cycle data sourced from public entities like NASA, CALCE, and ISU-ILCC. It allows for the reliable extraction of degradation indicators such as curvature, plateau length, and knee-related metrics, while minimizing the influence of dataset-specific discretization. The study tackles issues related to battery aging assessments stemming from variability among cells and differing cycling protocols. Significant correlations between knee onset and end-of-life were found across over 250 cells, with Pearson coefficients between 0.75 and 0.84. The framework enhances the identification of degradation transitions and improves early predictions of remaining useful life.
Key facts
- Framework analyzes lithium-ion battery aging using continuous trajectory representations
- Uses data from NASA, CALCE, and ISU-ILCC public datasets
- Examines more than 250 battery cells
- Shows Pearson correlations of 0.75-0.84 between knee onset and end-of-life
- Addresses cell-to-cell variability and heterogeneous cycling protocols
- Focuses on knee point identification and remaining useful life prediction
- Reduces sensitivity to dataset-specific discretization
- Extracts degradation descriptors including curvature and plateau length
Entities
Institutions
- NASA
- CALCE
- ISU-ILCC