精密光学透镜注射压缩成型多质量目标优化  

Multi-objective Optimization of Injection Compression Molding for Precision Optical Lens

作　　者：刘军辉[1,2] 陈新度 Liu Junhui;Chen Xindu(Mechanical and Electrical Engineering Institute,Heyuan Polytechnic,Heyuan 517000,China;Provincial Key Laboratory of Micro-nano Manufacturing,Guangdong University of Technology,Guangzhou 510006,China)

机构地区：[1]河源职业技术学院机电工程学院,广东河源517000 [2]广东工业大学,广东省微纳制造重点实验室,广州510006

出　　处：《工程塑料应用》2020年第7期56-60,共5页Engineering Plastics Application

基　　金：广东省特色创新项目(2018GKTSCX073);广东省科技计划项目(2017B090921007)。

摘　　要：为了提升精密光学透镜的综合质量,以粗糙度Ra与双折射数值φ为优化目标进行研究。首先针对传统的注射压缩模具结构缺陷,提出一种具有自动校准功能的注射压缩结构,之后以此为试验模具,提出一种包含BPNN建模、多目标优化智能算法NSGA-Ⅱ和试验验证的综合质量优化方法,以BPNN模型建立工艺参数与质量目标之间的关系,并以此作为NSGA-Ⅱ的适应度函数进行运算。试验结果表明,Ra和φ的BPNN模型具有很高的预测精度,其决定系数(R2)分别为0.97和0.94;通过对Pareto最优解与试验结果进行对比,2个质量目标的平均预测偏差都小于10%,优化预测与试验结果具有高度一致性,能够对产品的质量进行综合优化。In order to improve the comprehensive quality of the precision optical lens,the roughness Ra and the birefringenceφwere taken as the optimization objectives.Firstly,aiming at the defects of traditional injection compression mold structure,a kind of injection compression structure with automatic calibration function was proposed.Then,a comprehensive optimization method including BPNN modeling,multi-objective optimization intelligent algorithm NSGA-II and test verification was proposed.The BPNN models were used to establish the relationship between process parameters and objectives,which were used as the fitness function of NSGA-Ⅱ.The experimental results show that the BPNN models of Ra andφhave a high prediction accuracy,coefficient of determination(R2)of which are 0.97 and 0.94 respectively.Through the comparison of Pareto optimal solution and test results,the average prediction deviation of the two objectives are less than 10%.The optimization prediction is highly consistent with the test results,which can comprehensively optimize the product quality.

关 键 词：注射压缩成型 精密光学透镜 多目标优化 NSGA-Ⅱ算法 BP神经网络 

分 类 号：TQ320.66[化学工程—合成树脂塑料工业]