Machine learning‐based analysis and prediction of the interfacial corrosion processes of copper cathode plates during the electrolytic production of copper powders

Abstract Present efforts to support the essential industrial‐scale electrolytic production of copper‐based metal powders urgently require approaches to the real‐time predicting of corrosion of copper cathodes employed in electrolytic production processes. However, current approaches are extremely limited owing to the difficulty of accurately modeling the complex cathode corrosion process. In this study, the corrosion process under different parameters was analyzed by a self‐designed continuous electrolytic corrosion experimental device, clarify the influence mechanism of current density on the corrosion of the solid–liquid–gas interface area, and addresses this issue by applying a random forest machine learning approach based on three process parameters, including the electrolyte temperature, liquid‐level fluctuation cycle period, and current density. The dataset employed in the model is obtained using a novel experimental corrosion test method based on electrode arrays. The experimental results include the corrosion rates of copper cathode plates at different positions relative to the liquid electrolyte level during the electrolysis process. The resulting stochastic model is demonstrated to obtain a high prediction accuracy of 97% for the various regions of copper cathode plates defined according to liquid electrolyte level.