PhD Comprehensive Seminar | Ziyi Cai, Fast-Charging Lithium-Ion Batteries: Optimization with SPMeT and a Reinforcement Learning Approach

M3 3127

Candidate

Ziyi Cai | Applied Mathematics, University of Waterloo

Title

Fast-Charging Lithium-Ion Batteries: Optimization with SPMeT and a Reinforcement Learning Approach

Abstract

In this seminar, we will investigate a battery model - the Single Particle Model with Electrolyte and Thermal coupling (SPMeT) that is widely used in addressing the intricate challenge of optimizing lithium-ion battery charging. We will introduce and articulate a nonlinear optimal control problem with the objective of minimizing charge time while concurrently monitoring battery temperature and lifespan. This dual focus is pivotal for ensuring safety, as it necessitates adherence to strict operational constraints that prevent overheating and extend battery longevity. Central to our exploration is the utilization of a Reinforcement Learning framework to discover and refine charging policies. This approach leverages the flexibility and learning capabilities of RL to adapt to the battery's evolving conditions, thereby optimizing charging strategies in real-time.