Machine Learning

Solid-State Battery Li-Plating Dynamics Detection

Machine Learning

Solid-State Battery Li-Plating Dynamics Detection

A deep learning computer vision model for dynamic analysis of Li metal structures Researchers proposed and trained a machine learning (ML) computer vision-based auto-segmentation method (“batteryNET”), analyzing micro-computed tomography (μCT) datasets to study the dynamics of Li structures in Li-metal/polymer electrolyte batteries. The ML model semantic segmentation result demonstrates singular Li-related component changes, addressing diverse morphologies in the dataset. The visualizations of the cycled Li morphologies are provided below, including calculations about the volume and effective thickness of electrodes, deposited Li, and redeposited Li. This study discusses the spatial relationships between these components. The approach focuses on developing a computer vision ML model to detect and analyze the dynamics of lithium metal/polymer electrolyte battery cycling.

BatteryNET, a machine learning computer vision auto-segmentation system, was used to quantify dead Li, deposited, Li, and redeposited Li relative to the volume of the electrodes, allowing for quick and accurate analysis of lithium-metal battery datasets.

Academia | P. 229 Source: Huang, Y., Perlmutter, D., Fei-Huei Su, A. et al.

Extracted from the PDF; charts and images may read imperfectly.