【摘要】 利用多层前向神经网络 ,使用B P算法对结构钢回火性能预测进行了研究。通过利用 99种钢 4 5 0余组训练数据样本对神经网络进行训练 ,建立了结构钢回火后的力学性能与金属成分和回火温度之间的隐性函数。并针对训练用样本不足的问题 ,设计了为网络提供自学功能软件。在钢完全淬透的前提下 ,用此神经网络模型可在一定精度范围内预测结构钢的回火力学性能。更多还原

【Abstract】 In this paper,by using a back propagation(B P)algorithm,a feed forward network is applied to study the mechanical properties of structural steels after tempering.By training the neural network with the data of samples,mechanical properties of the steels after tempering as the function of its chemical components and tempering temperature are established.Aimed at the problem that the experiment lacks samples,a interface is designed to help the network to study by itself.With the help of this neural network model,the mechanical properties of the steels after tempering can be predicted in a certain precision.更多还原