Abstract
The performance of electrochemical cells for energy storage and conversion can be improved by optimizing their manufacturing processes. This can be time-consuming and costly with the traditional trial-and-error approaches. Machine Learning (ML) models can help to overcome these obstacles. In academic research laboratories, manufacturing dataset sizes can be small, while ML models typically require large amounts of data. In this work, we propose a simple but still novel application of a Transfer Learning (TL) approach to address these manufacturing problems with a small amount of data. We have tested this approach with pre-existing experimental and stochastically generated datasets. These datasets consisted of component properties (e.g., electrode density) related to different manufacturing parameters (e.g., solid content, comma gap, coating speed). We have demonstrated the robustness of our TL approach for manufacturing problems by achieving excellent prediction performance for electrodes in lithium-ion batteries and gas diffusion layers in fuel cells.