论文部分内容阅读
采用更为合理的建模参数,将预测变形镁合金力学性能的神经网络模型进行改进,并将此模型用于发展新型镁合金;对所有建模参数以全排列组合训练的方式构建模型,并通过比较这些模型的预测误差及相关系数来确定最合理的建模参数。模型的应用主要有Mg-Zn-Mn和Mg-Zn-Y-Zr两种合金。运用改进后的模型对Mg-Zn-Mn合金的力学性能进行预测,研究Mg-Zn-Y-Zr合金中Y/Zn摩尔比对强度的影响。最后,还利用此模型发展了一种高强挤压态的Mg-Zn-Y-Zr合金。结果表明:模型预测值与实验值吻合较好,改进后的模型可以用于发展新型变形镁合金。
A more reasonable modeling parameter is adopted to improve the neural network model which predicts the mechanical properties of the deformed magnesium alloy. The model is used to develop the new type of magnesium alloy. The model is constructed by all-permutation and combination training The most reasonable modeling parameters are determined by comparing the prediction errors and the correlation coefficients of these models. The main applications of the model are Mg-Zn-Mn and Mg-Zn-Y-Zr two alloys. The improved model was used to predict the mechanical properties of Mg-Zn-Mn alloys and the effect of Y / Zn molar ratio on the strength of Mg-Zn-Y-Zr alloys was investigated. Finally, a high-strength Mg-Zn-Y-Zr alloy was also developed by using this model. The results show that the predicted values agree well with the experimental ones, and the improved models can be used to develop new deformed magnesium alloys.