基于HsCAE和BP神经网络汽车饰件工艺优化  

Process Optimization of Automotive Trim Based on HsCAE and BP Neural Network

作　　者：马旭东[1] 孙理 栾冬[3] MA Xudong;SUN Li;LUAN Dong(Changchun Automobile Industry Institute,Changchun,Jilin 130013,China;FAW Jiefang Automotive Co.,Ltd.,Changchun,Jilin 130011,China;Harbin Institute of Technology,Weihai,Shandong 264209,China)

机构地区：[1]长春汽车工业高等专科学校,吉林长春130013 [2]一汽解放汽车有限公司,吉林长春130011 [3]哈尔滨工业大学,山东威海264209

出　　处：《塑料》2021年第2期87-92,98,共7页Plastics

摘　　要：结合车用饰件高质量的成型问题,对饰件注塑工艺参数进行了分析优化。首先,基于华塑CAE软件对塑件进行正交试验分析优化,选择翘曲、缩水和残余应力3个指标,结合各指标权重进行综合分析并评分,得到了最小综合评分为0.063。然后,利用正交试验样本建立了6-10-3的BP神经网络,通过训练获得一个拟合度较高网络模型。最后,利用遗传算法进行寻优,结合隶属度函数,完成了适应度函数的设计,将综合评分的最小值作为寻优目标,通过优化得到了饰件的最佳成型工艺参数,在CAE验证后,得到了综合评分值为0.031,与正交试验优化得到的综合评分相比,提升了50.8%,为未来汽车饰件的质量优化提供了参考。In order to solve the problem of high quality molding of automotive accessories,the injection molding process parameters were analyzed and optimized.First,based on the CAE software of Hua Su,the plastic parts were optimized by orthogonal test,and three indexes of warpage,shrinkage and residual stress were obtained.Combined with the weight of each index,the minimum comprehensive score was 0.063.Then,a 6-10-3 BP neural network was built with orthogonal test samples,and a network model with high fitting degree was obtained through training.Finally,genetic algorithm was used to optimize,and the fitness function was designed based on the membership function.The minimum comprehensive was taken as the optimization objective,and the optimal molding process parameters of the trim were optimized.After CAE verification,the comprehensive score was 0.031,which was 50.8%higher than the optimization of orthogonal test,providing reference for the quality optimization of automotive trim in the future.

关 键 词：华塑CAE BP神经网络 正交试验 遗传算法 注塑成型 

分 类 号：TQ320.66[化学工程—合成树脂塑料工业] TP391.7[自动化与计算机技术—计算机应用技术]