液态挤压工艺ANN/GA建模与优化研究

An ANN/GA(Artificial Neural Network/Genetic Algorithm) for Modeling and Optimizing of Liquid Metal Extrusion Process

利用人工神经网络方法 (ANN)建立了工艺系统模型 ,用遗传算法 (GA)对过程参数进行优化 ,实验结果与预测值吻合良好 ,为预测和控制该工艺成形质量提供了行之有效的手段。

Forming quality of liquid metal extrusion process has been difficult to ensure. We resolved this difficult problem by proposing modeling with artificial neural network (ANN) and optimization of process parameters with genetic algorithm (GA). Fig.1 shows the neural network model that can make the five process parameters selected compatible: (1) pouring temperature (T_1), (2) die temperature (T_2), (3) pressing velocity (v), (4) delaying period before applying pressure (t), (5) deforming force (F). Then we optimized the five parameters with GA and obtained T_1=716℃, T_2=250℃, t=30 s, v≈2.6×10 -3 m/s, F_ min=86.6 MPa for a liquid AlCuSiMg alloy extrusion. These predicted optimal values agreed well with test results.