中南大学学报（自然科学版）2025，Vol.56Issue(3)：881-890,10.DOI:10.11817/j.issn.1672-7207.2025.03.006
基于神经网络的6063铝型材挤压工艺多目标优化
Multi-objective optimization of extrusion process for 6063 aluminum profile based on neural network
摘要
Abstract
The metallographic analysis of a typical 6063 aluminum alloy extrusion profile was carried out.The extrusion process of the profile was simulated by finite element numerical simulation method.In order to improve the problem of uneven microstructure caused by uneven temperature in the cross section of 6063 aluminum profile outlet,an optimization method based on numerical simulation and neural network was proposed.Based on the GABP neural network,the mapping relationship between process parameters(extrusion speed,billet temperature,mold temperature and extrusion barrel temperature)and forming quality(average temperature Tav and temperature standard deviation DSDT of the cross section of the profile outlet)were established.Then,based on the NSGA-Ⅱ algorithm and Matlab software platform,the extrusion process parameters were optimized,and the better combination of process parameters was obtained.The results show that there are obvious differences in the grain structure of different parts.The inhomogeneity of the profile structure is mainly caused by the inhomogeneity of the profile temperature at the extrusion scheme.The better combination of process parameters are as follows.The extrusion speed is 3.73 mm/s,the billet temperature is 474.1℃,the mold preheating temperature is 469.9℃,and the extrusion barrel temperature is 456.8℃.Compared with the initial extrusion scheme,the DSDT decreases from 5.33℃to 3.32℃after extrusion with optimized parameters.Finally,this set of optimal process parameters are tested and trial-produced,and the grain structure uniformity of the profile is effectively improved.关键词
6063铝型材/组织均匀性/GABP神经网络/NSGA-Ⅱ算法/Qform软件Key words
6063 aluminum profile/microstructural uniformity/GABP neural network/NSGA-Ⅱ algorithm/Qform software分类
金属材料引用本文复制引用
刘鹏程,彭炳锋,刘寒龙,刘莹雪,孙立科,林高用..基于神经网络的6063铝型材挤压工艺多目标优化[J].中南大学学报（自然科学版）,2025,56(3):881-890,10.基金项目
国家重点研发计划项目(2022YFC3901702)(Project(2022YFC3901702)supported by the National Key Research and Development Program of China) （2022YFC3901702）
