论文部分内容阅读
介绍了一种神经网络法的计算机信息工艺,它能在不同温度和不同变性剂含量的条件下,准确地预测出焊缝的若干抗裂性指标。通过神经网络法预测值与实际试验值的对比发现,变性剂中w(Ce)在0.025%～0.040%范围时,所得到的抗裂性指标处于最佳状态;当w(Ce)高于0.040%后,指标值开始下降。采用变性剂Ce进行焊缝金属的合金化,能够细化焊缝组织,净化晶粒边界,从而增强焊缝的抗裂性能。
A computerized information technology of neural network method is introduced, which can accurately predict some crack resistance indexes of weld under different temperature and different denaturing agent content. The comparison between the predictive value and the actual experimental value by neural network shows that the best anti-cracking index is obtained when w (Ce) is in the range of 0.025% ~ 0.040% %, The index value began to decline. The use of a denaturing agent Ce for alloying of the weld metal can refine the weld microstructure and clean the grain boundaries to enhance the crack resistance of the weld.