摘要
在实验数据的基础上,采用附加动量项和变步长的方法,对人工神经网络的BP算法进行了训练。利用训练后所得到的模型,对屈服强度进行了分析和预测。计算表明,网络预测值与实测值之间具有很高的相关性和精确度,为屈服强度提供了一定的理论辅助手段。
Based on experimental data, the back-propagating algorithm in artificial neural networks （ANNs） was trained by appending momentum and changing steps. According to the trained model, the critical points for yield strength were analyzed and predicted. The results show that the prediction precision and the pertinency between the predicted ANNs and measured values are considerably high. A Theoretical method for prediction martensite start temperature is given,
出处
《热加工工艺》
CSCD
北大核心
2006年第10期58-59,共2页
Hot Working Technology
关键词
屈服强度
神经网络
BP算法
yield strength
neural networks
back-propagating
