RSM结合GA-BP的聚碳酸酯细长杆轴向变形预测

Polycarbonate Slender Pen Rod Axial Deformation Prediction Combined RSM with GA-BP

对聚碳酸酯(PC)注塑的4个工艺变量的遗传算法-BP网络(GA-BP)和响应面(RSM)2种优化方法进行了比较,并综合应用2种方法的优点,优化预测了PC细长杆填充过程中的工艺参数。为获得最小的轴向变形,选择模具温度、熔体温度、保压时间、保压压力4个工艺变量,通过RSM获得最优的工艺参数组合为模具温度85℃,熔体温度290℃,保压时间7s,保压压力60MPa,应用GA-BP预测此工艺条件下的最小的轴向变形值为0.0064mm,Moldflow软件的模拟值为0.0063mm,预测精度达到1.58%;通过与9组随机试验的模拟精度比较,GA-BP和RSM的预测最大绝对误差分别为5.48%、7.41%,相对误差分别为2.76%、3.13%,通过GA-BP和RSM二者联合预测精度有了很大提高。

Four process parameters, mold temperature, melt temperature, holding time, and holding pressure, in injection molding of slender polycarbonate（PC） parts were optimized using GA-BP and RSM methods. The RSM method resulted in the optimal parameter combination of mold temperature 85℃, melt temperature 290℃, holding time 7 s, holding pressure 60 MPa. The GA-BP method predicted that the minimum axial deformation was 0. 0064 mm, very close to the Moldflow simulated value of 0. 0063 mm. For 9 groups of random trials, the maximum absolute errors of prediction by GA-BP and RSM were 5.48℃ and 7.41%, and the relative errors were 2.76% and 3.13%, respectively. It showed that the combination of GA-BP and RSM provided a high prediction accuracy.