Advisor: Dr. Chao Hu (Major Professor), Dr. Zhaoyu Wang (Co-major Professor)
Title: Li-ion Battery Capacity-Trajectory Prediction Using Early-Life Data
Abstract: It is challenging to predict the capacity trajectory of a lithium-ion (Li-ion) battery cell when the cell shows negligible capacity fade. We propose an end-to-end learning framework that closely couples empirical capacity fade models and data-driven machine learning models. A critical element of our end-to-end learning framework is an end-to-end learning problem formulated to simultaneously fit a selected empirical model to estimate the capacity trajectory and train a machine learning model to estimate the parameters of the empirical model using early-life data (e.g., voltage and current measurements from the first 100 cycles). We demonstrate this learning framework using a publicly available dataset of 169 lithium-iron-phosphate/graphite battery cells cycled under fast-charging protocols. Our best model achieves similar performance in cycle life prediction compared with existing early life prediction models. More importantly, our proposed end-to-end learning framework can predict the entire capacity trajectory. We further extend the framework’s capabilities to uncertainty quantification by estimating the associated confidence interval over the entire trajectory.
Biography: Jinqiang Liu is a Ph.D. candidate in Electrical Engineering at Iowa State University. His research focuses on data-driven lithium-ion battery degradation modeling and embedding machine learning models inside closed-loop decision-making processes.