AI Modeling of Roll-to-Roll Coating System for Manufacturing Secondary Battery Electrodes

Abstract

Innovation in artificial intelligence (AI) has rapidly expanded its applicability across various industrial processes, demonstrating particularly strong effectiveness in continuous manufacturing systems such as roll-to-roll (R2R) production. Secondary battery manufacturing relies heavily on R2R-based electrode fabrication, where slight variations in tension, line speed, or coating conditions can significantly influence the final product quality. In this study, an automated data pipeline was developed to support AI modeling of the electrode manufacturing process for secondary batteries. The proposed pipeline consists of data synchronization, region partitioning, file fusion, and automated input variable selection. Based on this pipeline, key process variables required for mass loading prediction were systematically identified and optimized. This study proposes a data-handling workflow for battery-manufacturing process data and evaluates an AI-based predictor on five continuous-production cases. On the hold-out test sets, the models achieved a maximum R² of 0.966, and four of the five cases attained R² values of at least 0.95. The RMSE values ranged from 20.125 to 52.097.