基于ACO-BP神经网络的储能电源上壳件气辅成型工艺参数优化  

Optimization of gas-assisted molding process parameters for upper shell of energy storage power supply based on ACO-BP neural network

作　　者：杨明 刘赛 刘巨保[1] 李峰 姚建锋 刘厚德 YANG Ming;LIU Sai;LIU Jubao;LI Feng;YAO Jianfeng;LIU Houde(School of Mechanical Science and Engineering,Northeast Petroleum University,Daqing 163318,China;Hongfeng Intelligent Manufacturing(Shenzhen)Co.,Ltd.,Shenzhen 518117,China)

机构地区：[1]东北石油大学机械科学与工程学院,黑龙江大庆163318 [2]弘丰智能制造(深圳)有限公司,广东深圳518117

出　　处：《工程塑料应用》2025年第4期101-107,共7页Engineering Plastics Application

摘　　要：以某储能电源上壳件为研究对象,为提高气辅成型效果,运用Moldflow软件进行气辅成型数值模拟。以气体穿透体积和最大翘曲变形量为优化指标设计了正交试验,利用Critic权重法确定权重占比,通过计算综合评分将双目标优化转化为单目标优化;建立工艺参数与综合评分之间的BP神经网络模型,利用蚁群算法(ACO)进行全局寻优。结果表明,当熔体预注射量为93%、熔体温度为270℃、模具温度为87.78℃、延迟时间为3.5 s、气体压力为35 MPa、气体注射时间为20 s、冷却时间为179.62 s时,综合评分值最大、工艺方案最优。利用Moldflow软件对最优工艺参数进行验证,结果显示气体穿透体积为10.6975%、最大翘曲变形量为2.169 mm,计算其线性组合得到综合评分为1.0495,与优化算法结果的误差仅为1.1%,并进行试模验证,试模翘曲结果与模流分析结果误差为2.8%,且产品无吹穿、吹破等缺陷,表面质量良好。以上研究结果表明,基于ACO-BP神经网络优化气辅成型工艺参数的技术方法具有可行性。Taking the upper shell of an energy-storage power supply as the research object,in order to improve the gas-assisted molding effect,the Moldflow software was used for numerical simulation of gas-assisted molding.An orthogonal experiment was designed,with the gas penetration volume and total warpage deformation as the optimized indexes.The Critic weight method was utilized to determine the weight proportion,and the dual-objective optimization was transformed into single-objective optimization by calculating the comprehensive score.The BP neural network model was established between the process parameters and the comprehensive score,and the ant colony algorithm(ACO)was used for global optimization.The results show that when the melt pre-injection volume is 93%,the melt temperature is 270℃,the mold temperature is 87.78℃,the delay time is 3.5 s,the gas pressure is 35 MPa,the gas injection time is 20 s,and the cooling time is 179.62 s,the comprehensive score is the largest,and the process plan is the optimal.The process parameters obtained by optimization were verified using the Moldflow software.The results show that the gas penetration volume is 10.6975%and the total warpage deformation is 2.169 mm.Calculating their linear combination gives a comprehensive score of 1.0495,with an error of only 1.1%compared with the result of the optimization algorithm.The trial-mold verification was also carried out.The error between the warpage results of the trial-mold and the result of the mold-flow analysis is 2.8%,and the product has no defects such as blow-through and burst,with good surface quality.The above research indicates that the technical method of optimizing gas-assisted molding process parameters based on the ACO-BP neural network is feasible.

关 键 词：气体辅助注射成型 数值模拟 参数优化 BP神经网络 蚁群算法 

分 类 号：TQ320.66[化学工程—合成树脂塑料工业]