Current Efficiency of Low Temperature Aluminum Electrolysis Studied by Neural Network

A prediction model for Current Efficiency (CE) of low temperature aluminum electrolysis (LTAE) with the low molar ratio electrolyte of Na₃AIF₆-AIF₃-CaF₂-MgF₂-LiF-Al₂O₃ system was investigated based on artificial neural network principles. The nonlinear mapping between CE of LATE and various electrolytic conditions was obtained from a number of experimental data and used to predictCE of LATE. The trained neural networks possessed high precision and resulted in a good predicting effect. As a result, artificial neural networks as a new cooperating and predicting technology provide a new approach to the further studies on low temperature aluminum electrolysis.