基于径向基神经网络的导光条注塑工艺优化  

Radial Basis Neural Network-Based Light Guide Strip Injection Molding Process Optimization

作　　者：苏通 黄瑶[1] 倪君杰 魏翔宇 Su Tong;Huang Yao;Ni Junjie;Wei Xiangyu(School of Material Science and Engineering,Jiangsu University,Zhengjiang,Jiangsu,212013;Zhengjiang Success Group Co.,Ltd.,Zhengjiang,Jiangsu,212028)

机构地区：[1]江苏大学材料科学与工程学院,江苏镇江212013 [2]镇江成功高科技有限公司,江苏镇江212028

出　　处：《现代塑料加工应用》2021年第1期36-39,共4页Modern Plastics Processing and Applications

摘　　要：以汽车前组合灯导光条为例,选择最优拉丁超立方抽样方法得到样本。选取熔体温度、模具温度、保压时间、保压压力和冷却时间5个参数为输入层,以最小体积收缩率与最小缩痕指数为输出层,构建径向基(RBF)神经网络模型。建立的模型经检验,拟合度高,误差小,可以替代仿真程序。应用Isight优化模块,得到一组最优注塑工艺参数组合,实际模拟结果和预测结果基本吻合,有效提高了成型质量。Taking the light guide strip of automobile front combination lamp of the car as an example, the optimal Latin hypercube sampling method was selected to obtain samples. The five parameters of melt temperature, mold temperature, holding time, holding pressure and cooling time were selected as the input layer, and the smallest volume shrinkage rate and the smallest sink mark index were used as the output layer to construct a RBF neural network model. The established model has been tested and has a high degree of fit and small error, which can replace the simulation program. The Isight optimization module was used to obtain a set of optimal injection molding process parameter combinations. The actual value is basically consistent with the predicted value, which effectively improves the molding quality.

关 键 词：导光条 注塑工艺 径向基神经网络 最优拉丁超立方 参数优化 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TQ320.662[自动化与计算机技术—控制科学与工程]