摘要
建立了以纯金属原子半径、熔点、沸点和原子化焓预测表面张力的人工神经网络模型.训练后的神经网络能较好的拟会实验数据.对40种金属的表面张力进行回想和预测结果与实验值的偏差在可接受范围内,表明人工神经网络在纯金属表面张力预测方面有一定的前景.
An artificial neural network was established to forecast the surface tension of pure metal from the experimental data of atomic radius, melting point, boiling point and atomization enthalpies. The trained network can represent the relahonship between the input factors and output factor (surface tension).The associated and forecast data for more than 40 pure metals are acceptable considering the deviation of the experimental dara for surface tension, which shows a good prospect of artificial neural network in the predic-tion of surface tension of pure metals.
出处
《北京科技大学学报》
EI
CAS
CSCD
北大核心
1997年第3期287-290,共4页
Journal of University of Science and Technology Beijing
基金
国家自然科学基金
关键词
表面张力
神经网络
反传学习网
金属
预测
surface tension, artificial neural network, back propagation learning