摘要
针对塑料模型注塑成型优化过程中工艺参数多、计算准确度低、工程模拟量大的特点。以塑料齿轮零件为例,通过引入BP神经网络技术,结合Moldflow软件建立注塑成型工艺参数优化模型。以体积收缩率和翘曲变形量为注塑工艺评定目标函数,选择熔体温度、保压压力、保压时间、模具表面温度为训练样本,建立44正交试验表,由相对方差分析评价模型的分析结果,给出优化后的工艺参数,指导工程实际应用。研究结果表明,通过BP神经网络对初始工艺参数进行训练,模型训练预测值与模拟值相对误差在3%以下,满足预测精度要求,经过对正交试验表样本进行训练,确定优化工艺参数为:熔体温度220℃、保压压力50 MPa、保压时间15 s,模具温度70℃。由Moldflow模型验证指出优化后的工艺参数组合能减少塑料件的注塑缺陷,提升塑料件的使用性能。
In the process of plastic model injection molding optimization,the process parameters are many,the calculation accuracy is low,and the engineering simulation quantity is large.Taking plastic gear as an example,the optimization model of injection molding process parameters is established by introducing BP neural network technology and combining Moldfow software.Taking volume shrinkage and warp deformation as the objective function of injection molding process evaluation,the solution temperature,pressure holding pressure,pressure holding time and mold surface temperature were selected as training parameters.The 44 orthogonal test table was established.The analysis results of the relative analysis of variance(ANOVA)evaluation model are obtained,and the optimized process parameters are given to guide the practical application of the project.The results show that the initial process parameters are trained by BP neural network,the relative error between the model training prediction value and the simulation value is less than 3%,which meets the requirements of prediction accuracy.The optimized process parameters are determined as follows:solution temperature 220℃,holding pressure 50 MPa,holding time 15 s,die temperature 70℃.According to the Moldfow model,the optimized process parameter combination can reduce the injection defects and improve the performance of plastic parts.
作者
项丽萍
杨红菊
Xiang Liping;Yang Hongju(Department of Information Engineering,Jincheng Institute of Technology,Jincheng 048000,China;School of Computer and Information,Shanxi University,Taiyuan 030006,China)
出处
《工程塑料应用》
CAS
CSCD
北大核心
2021年第5期92-96,107,共6页
Engineering Plastics Application
基金
山西省“十二五”规划课题项目(GH-11141)。
关键词
塑料件注塑
神经网络
参数优化
plastic injection molding
neural network
parameter optimization
作者简介
通信作者:项丽萍,硕士,副教授,主要研究领域为数据库技术、云计算、大数据等,E-mail:muyuxuan066@163.com。
