基于神经网络的AZ31镁合金热挤压成形参数的优化

Optimization of Hot Extrusion Forming Parameters of AZ31 Magnesium Alloy Based on Neural Network

利用神经网络对AZ31镁合金热挤压成形工艺参数进行了优化。首先,根据正交试验获得了不同挤压比、挤压温度和挤压速度下AZ31镁合金试件的力学性能,并获得成熟的GA-BP神经网络。然后,利用该网络对工艺参数进行反向寻优,利用遗传算法的优化作用对工艺参数进行寻优。最后,找到了最佳工艺参数为挤压比23,挤压温度365℃,挤压速度2.6mm/min,对应的抗拉强度为296MPa。通过试验值与仿真值的比较得出本文构建的优化模型有较高精度。

The process parameters of hot extrusion forming of AZ31 magnesium alloy were optimized .by using neural network.Firstly,the mechanical properties of AZ31 magnesium alloy specimen under different extrusion ratios,extrusion temperatures and extrusion speeds were obtained according to the orthogonal experiment ,and a mature GA-BP neural network was obtained.Then,the network was used to optimize the process parameters,and the genetic algorithm was used to optimize the process parameters.Finally,it is found that the optimum technological parameters are extrusion ratio of 23, extrusion temperature of 365℃,and extrusion speed of 2.6mm/min.The corresponding tensile strength is 296MPa.Through the comparison of the test value and the simulation value,it is concluded that the optimization model constructed in this paper has high accuracy.