基于CAE和BP网络的注射成型质量预测

Prediction of the Injection Quality Based on CAE and BP Neural Network

在数值仿真和实验设计方法获得的样本数据基础上,利用BP网络在并行计算和数值逼近上的优势,获得了工艺参数与各质量指标的非线性映射关系,进而实现了在工艺因素变化情况下能快速、方便地预测出注射成型的质量。研究表明,在误差许可的范围内,经训练的神经网络模型可以实现对制品质量的正确预测,预测模型也可直接用于指导企业的在线生产或优化。

Based on the sample data obtained from numerical simulation and DOE method,BP netˉwork is used to acquire the non-linear mapping relations between the technological variables and the quality indices in virtue of its advantages in parallel computation and numerical approach.So it can help to easily and rapidly predict the part injection quality with the parameters varying.The reˉsearch shows that the neural network can make correct prediction within the range of the allowed tolerances and the prediction model can be used as a guidance for the on-line production or optiˉmization of enterprises.