Learning, Estimation, and Control for Batteries

This talk focuses on learning, estimation, and control methods to enhance electrochemical battery performance. We begin by providing an overview of mathematical models for batteries and the key challenges that motivate new paradigms for learning, estimation and control. Then we focus on “hybrid” models that combine physics with data-driven approaches. Next, we examine the challenge of predicting the remaining energy within a battery, a type of state estimation and prediction problem. The third part focuses on utilizing reinforcement learning for fast charging batteries. Finally, we close the talk by showcasing how machine learning can be utilized to predict battery health degradation.

Speaker: Scott Moura, UC Berkeley