神经网络预报低温铝电解的电流效率

NEURAL NETWORK PREDICTS CURRENT EFFICIENCY OF LOW TEMPERATURE ALUMINIUM ELECTROLYSIS

用Ｎａ３ＡｌＦ６－ＡｌＦ３－ＣａＦ２－ＭｇＦ２－ＬｉＦ－Ａｌ２Ｏ３体系中的低摩尔比（ｎ（ＮａＦ）／ｎ（ＡｌＦ３））电解质进行低温电解实验，以大量的实验数据为学习样本，让网络学习以建立各种电解条件与对应电流效率间的非线性映射，用来预报不同电解条件下的电流效率，研究了各种电解工艺参数对电流效率的影响规律。

This paper presents a new approach to predict current efficiency (CE) of low temperature aluminium electrolysis (LTAE) with the low molar ratio electrolyte of Na 3AlF 6-AlF 3-CaF 2-MgF 2-LiF-Al 2O 3 system. The nonlinear mapping between CE of LTAE and various electrolytic conditions is established according to a number of experimental data to predict CE of LTAE. The effect laws of various electrolytic parameters on CE are also given in the paper.