Leaning Governing Relations in Battery Electrodes: Hybridizing Physics- and Data-Driven Approaches

The vast design space and long lifetime of batteries have limited the speed of innovations at the materials, cell, and systems level. In this talk, I will overview efforts at Stanford to dramatically accelerate the pace of research and development for lithium-ion batteries by hybridizing physics-based and data-driven approaches. Specifically, I will present our recent work on predicting not only battery lifetime but also aging mechanisms, which lays the foundation for transferable learnings between multiple battery chemistries. I will also showcase new methods for learning heterogeneities at the particle and electrode levels by combining multi-fidelity characterizations, physics-based modeling, and machine learning.