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

作　　者：刘鹏程 彭炳锋 刘寒龙 刘莹雪 孙立科 林高用[1] LIU Pengcheng;PENG Bingfeng;LIU Hanlong;LIU Yingxue;SUN Like;LIN Gaoyong(School of Materials Science and Engineering,Central South University,Changsha 410083,China;Jiangxi Baotai Nonferrous Metals Group Co.Ltd.,Yingtan 335001,China;Beijing Intelligent United Innovation Technology Co.Ltd.,Beijing 100027,China)

机构地区：[1]中南大学材料科学与工程学院,湖南长沙410083 [2]江西保太有色金属集团有限公司,江西鹰潭335001 [3]北京创联智软科技有限公司,北京100027

出　　处：《中南大学学报(自然科学版)》2025年第3期881-890,共10页Journal of Central South University:Science and Technology

基　　金：国家重点研发计划项目(2022YFC3901702)。

摘　　要：对1种典型6063铝合金挤压型材金相进行分析。采用有限元数值模拟方法对该型材的挤压过程进行模拟。为了解决6063铝型材横截面组织不均匀问题,提出一种基于数值模拟和神经网络相结合的挤压温度均匀性多目标优化方法。基于GABP神经网络建立工艺参数(挤压速度、坯料温度、模具温度和挤压筒温度)和成形质量(型材出口横截面的平均温度T_(av)和温度标准差D_(SDT))的映射关系,基于NSGA-Ⅱ算法和Matlab软件平台,对挤压工艺参数进行优化,获得较佳的工艺参数组合。研究结果表明:铝型材不同区域的晶粒组织存在较明显差异;型材组织的不均匀性主要是挤出模口的型材温度不均匀所致;较佳的工艺参数组合是挤压速度为3.73mm/s、坯料温度为474.1℃、模具预热温度为469.9℃、挤压筒预热温度为456.8℃;与初始挤压工艺方案对比,采用优化的挤压工艺参数时,温度标准差DSDT从5.33℃下降到3.32℃。将这组最优工艺参数进行挤压生产验证,发现不同部位晶粒组织的均匀性大幅度提高。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 T_(av) and temperature standard deviation D_(SDT) 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 

分 类 号：TG379[金属学及工艺—金属压力加工]