应用BP网络建立新型高强韧钢的结构形成因子与抗拉强度的关系

Establishment of Relationship between Structure Formation Factors and Tensile Strength of a New High Strength-Toughness Steel by BP Network

根据固体与分子经验电子理论计算了 19组新型高强韧钢的试验数据所对应的结构形成因子 S值 ,从中随机抽取 16组数据 ,应用人工神经网络的 BP结构进行训练 ,用其余 3组数据进行检验 ,其平均训练误差为 0 .8731% ,平均检验误差为 0 .96 5 7%。建立了价电子结构与强度之间的定量关系 。

The structure formation factors S were calculated from 19 groups of experimental data of a new high strength toughness steels on the basis of the empirical electrionic theory of solids and molecules, and 16 groups of which selected randomly were applied to train with BP network of artificial neural networks and other three groups were used to examine. The average trained error is 0.873 1 %, and the average examined error is 0.965 7 %. The quantitative relationship between the valence electron structure and the tensile strength was built, and composition design of alloy was tried on the electron level.