摘要
基于BP(反向传播)神经网络理论,利用MATLAB R2012b软件的人工神经网络工具箱,以实验中得到的25组不同温度、时间、pH、硫酸镍浓度、次磷酸钠浓度、辅助配位荆浓度下的T10A钢针织器材化学镀镍层厚度及孔隙率为样本,对建立的BP神经网络模型进行训练及预测。结果表明,该BP神经网络有较快的学习速度及较高的预测精度。
A BP （back propagation） network model was established based on the ANN （artificial neural network） theory and then trained using the Matlab R2012b ANN toolbox via 25 groups of coating thickness and porosity data obtained by electroless plating nickel on T10A steel needle under different temperatures, time, pHs, and concentrations of nickel sulfate, sodium hypophosphite, and auxiliary complexing agent. The results indicated that the model learns fast and can be applied to prediction with relatively high accuracy.
出处
《电镀与涂饰》
CAS
CSCD
北大核心
2014年第4期150-154,共5页
Electroplating & Finishing
关键词
化学镀镍
反向传播网络
厚度
孔隙率
训练
预测
electroless nickel plating
back propagationnetwork
thickness
porosity
training
prediction
