摘要
通过对4Cr5MoSiV1模具钢分别进行不同的加速电压、照射距离和轰击次数下的电子束表面改性试验,取得模具钢试样的实际耐磨性测试数据,并将其作为神经网络的训练样本和验证样本,建立了3×12×1三层网络模型进行模具钢电子束表面改性的神经网络预测,并对网络的预测精度进行分析。结果表明,该三层神经网络可进行较高精度的模具钢电子束表面改性神经网络预测。
The wear resistance of die steel 4Cr5MoSiV1 with surface modification by using electron beam at different accelerating voltages, radiation distance and hulling index was measured. And their sample data were used for training and certification of the neural network. The neural network model with 3 ×12 ×1 three layers was established to predict the performance of surface modification die steel by electron beam, and the prediction precision of network model was discussed and researched. The results show that the neural network with three layers can be used for the good prediction of surface modification die steel by electron beam.
出处
《热加工工艺》
CSCD
北大核心
2013年第16期155-157,共3页
Hot Working Technology
关键词
神经网络
模具钢
电子束
表面改性
预测
neural network
die steel
electron beam
surface modification
prediction
