Please wait a minute...
金属学报  2017, Vol. 53 Issue (9): 1125-1132    DOI: 10.11900/0412.1961.2016.00573
  本期目录 | 过刊浏览 |
基于GA-ELM的铝合金压铸件晶粒尺寸预测
梅益1(), 孙全龙1, 喻丽华1, 王传荣2, 肖华强1
1 贵州大学机械工程学院 贵阳 550025
2 中国石油新疆独山子石化分公司 克拉玛依 833699
Grain Size Prediction of Aluminum Alloy Dies Castings Based on GA-ELM
Yi MEI1(), Quanlong SUN1, Lihua YU1, Chuanrong WANG2, Huaqiang XIAO1
1 College of Mechanical Engineering, Guizhou University, Guiyang 550025, China
2 China Petroleum Xinjiang Dushanzi Petrochemical Corp., Kelamayi 833699, China
全文: PDF(3139 KB)   HTML
  
摘要: 

为提高铝合金压铸件晶粒尺寸预测的效率和准确率,应用遗传算法-极限学习机(GA-ELM)模型预测晶粒尺寸。ELM的输入层权值矩阵及隐含层阈值矩阵具有随机性,通过GA算法对ELM的输入层权值矩阵和隐含层阈值矩阵进行优化,建立GA-ELM模型。以晶粒尺寸作为输出参数,相关压铸工艺参数作为输入参数,通过压铸生产实验及金相测量获得相应数据,对GA-ELM模型进行实例分析,并与同样使用遗传算法优化的GA-BP神经网络模型和原始ELM模型预测结果进行对比。最后,通过金相组织测量实验验证GA-ELM模型预测结果的可靠性。结果表明,利用GA-ELM模型预测铝合金压铸件晶粒尺寸具有较高的预测精度及预测效率,与其它算法相比,具有一定的优越性。

关键词 铝合金微观组织晶粒尺寸极限学习机GA-ELM模型    
Abstract

Effective grain size prediction for aluminum alloy die castings is of great significance to the rational formulation of die casting process parameters and to the improvement of casting mechanical properties. The traditional grain size prediction method cannot give consideration to both the efficiency and accuracy because of its inherent defects. To improve the efficiency and accuracy of predicting grain size for aluminum alloy die castings, this study proposes a prediction method that is based on the genetic algorithm-extreme learning machine (GA-ELM) model. ELM has the characteristics of few parameter settings, fast learning and good generalization performance, but the algorithm randomly generates the initial input layer weight matrix and the hidden layer threshold matrix, which greatly affects the prediction result. By exploiting GA's excellent global optimization ability, the optimal initial input layer weight matrix and the hidden layer threshold matrix for ELM can be found. The establishment of GA-ELM model can considerably improve the prediction accuracy of ELM model. This study uses grain size as the output parameters and relevant die casting process parameters as the input parameters. The castings produced under different die-casting process parameters are obtained experimentally, and the microstructures of specified sections of key casting positions are analyzed and measured to obtain the average grain size of the sec tions, i.e. the output parameters. The GA-ELM model is trained and tested using these data. To verify the superiority of the GA-ELM model in grain size prediction, this study compares the prediction results of GA-ELM model with the GA-BP neural network model and the original ELM model, and eventually verifies the reliability of GA-ELM model prediction results through metallographic structure measurement experiment. The results show that the GA-ELM model has higher prediction accuracy than the GA-BP neural network model and the original ELM model. Besides, its prediction efficiency is higher than the GA-BP model, while is lower than the original ELM model. With fairly high prediction accuracy and efficiency, the GA-ELM model can meet the actual engineering requirements. Furthermore, its prediction reliability and excellent prediction effect are verified by the results of metallographic structure measurement experiment.

Key wordsaluminum alloy    microstructure    grain size    extreme learning machine (ELM)    GA-ELM model
收稿日期: 2016-12-27     
ZTFLH:  TG146.2  
基金资助:贵州省科学技术基金项目No.20142053
作者简介:

作者简介 梅益,男,1974年生,教授,博士

引用本文:

梅益, 孙全龙, 喻丽华, 王传荣, 肖华强. 基于GA-ELM的铝合金压铸件晶粒尺寸预测[J]. 金属学报, 2017, 53(9): 1125-1132.
Yi MEI, Quanlong SUN, Lihua YU, Chuanrong WANG, Huaqiang XIAO. Grain Size Prediction of Aluminum Alloy Dies Castings Based on GA-ELM. Acta Metall Sin, 2017, 53(9): 1125-1132.

链接本文:

https://www.ams.org.cn/CN/10.11900/0412.1961.2016.00573      或      https://www.ams.org.cn/CN/Y2017/V53/I9/1125

图1  遗传算法-极限学习机(GA-ELM)算法流程
图2  汽车空压机端盖压铸件几何模型
No. TP / ℃ TI / ℃ VS / (ms-1) VF / (ms-1)
1 190 650 0.2 3
2 200 660 0.3 4
3 210 670 0.4 5
4 220 680 0.5 6
表1  各成型工艺参数及水平设置
图3  汽车空压机端盖接口部位几何模型
No. TP / ℃ TI / ℃ VS / (ms-1) VF / (ms-1) D / μm
1 650 190 0.2 3 250
2 650 200 0.3 4 349
3 650 210 0.4 5 282
4 650 220 0.5 6 303
5 660 190 0.3 5 282
6 660 200 0.2 6 251
7 660 210 0.5 3 313
8 660 220 0.4 4 328
9 670 190 0.4 6 274
10 670 200 0.5 5 351
11 670 210 0.3 4 346
12 670 220 0.2 3 251
13 680 190 0.5 4 245
14 680 200 0.4 3 259
15 680 210 0.3 6 245
16 680 220 0.2 5 280
17 650 190 0.5 3 260
18 650 200 0.4 4 281
19 650 210 0.3 5 294
20 650 220 0.2 6 277
21 660 190 0.5 4 320
22 660 200 0.4 3 227
23 660 210 0.3 6 288
24 660 220 0.2 5 379
25 670 190 0.4 5 268
26 670 200 0.5 5 301
27 670 210 0.2 3 292
28 670 220 0.3 4 320
29 680 190 0.4 6 213
30 680 200 0.5 5 216
31 680 210 0.2 4 280
32 680 220 0.3 3 246
表2  训练集工艺参数实验结果
No. Parameter Setting
1 Number of ELM hidden layer nodes 32
2 Population size 60
3 Maximum iterations 150
4 Crossover possibility 0.8
5 Mutation probability 0.5
6 Objective function The minimum of average relative errors
7 Fitness evaluation method Linear evaluation
8 Value range of weights Wij [-1, 1]
9 Value range of thresholds bi [-1, 1]
10 Terminal condition Maximum number of iterations
表3  GA-ELM参数设置表
No. TP / ℃ TI / ℃ VS / (ms-1) VF / (ms-1) D / μm
1 650 200 0.4 6 271
2 650 210 0.5 5 308
3 650 210 0.5 6 258
4 660 200 0.3 4 270
5 660 210 0.4 3 235
6 670 190 0.5 5 262
表4  测试集工艺参数实验结果
图4  遗传算法进化结果
图5  GA-ELM模型对测试集样本的预测结果与测试集输出真值对比
图6  GA-BP模型对测试集样本的预测结果与测试集输出真值对比
图7  ELM模型对测试集样本的预测结果与测试集输出真值对比
Model index Ea / μm Er / % t / s
GA-ELM 10.2 3.8 47.03
GA-BP 14.4 5.5 107.02
ELM 22.8 8.6 2.23
表5  3种预测模型预测性能对比
图8  汽车空压机端盖接口部位同一截面不同位置的金相组织
[1] Qin P C, Zhang X J.Application of Procast software in die casting's numerical simulation[J]. Hot. Work. Technol., 2010, 39(23): 75(秦鹏程, 张希俊. Procast在压力铸造数值模拟的应用现状[J]. 热加工工艺, 2010, 39(23): 75)
[2] Li S Z.Numerical simulation and processing optimization of large complex aluminum die-casting [D]. Wuhan: Huazhong University of Science and Technology, 2012(李世钊. 大复杂铝合金压铸件成数值模拟及工艺优化 [D]. 武汉: 华中科技大学, 2012)
[3] Xiong M H.CAE technology on filling solidification processes and die casting process of aluminum alloy die casting[J]. Hot. Work. Technol., 2013, 42(15): 63(熊明辉. 铝合金压铸件充型凝固过程及其压铸工艺CAE分析[J]. 热加工工艺, 2013, 42(15): 63)
[4] Yu B.Technology optimization of high press die cast aluminum alloy cylinder cover based on numerical simulation analysis[J]. Found. Technol., 2011, 32: 1109(于波. 基于数值模拟分析的铝合金缸盖罩压铸工艺优化[J]. 铸造技术, 2011, 32: 1109)
[5] Huang X F, Xie R, Tian Z Y, et al.The development and the outlook of the die casting Technology[J]. New Technol. New Proc., 2008, (7): 50(黄晓锋, 谢锐, 田载友等. 压铸技术的发展现状与展望[J]. 新技术新工艺, 2008, (7): 50)
[6] Zi B T, Yao K F, Cui J Z, et al.A study on the artificial neural network model of the solidified grain size of Al-alloy[J]. J. Appl. Sci., 2001, 19: 353(訾炳涛, 姚可夫, 崔建忠等. 铝合金凝固晶粒尺寸的人工神经网络研究[J]. 应用科学学报, 2001, 19: 353)
[7] Liu B, Tang A T, Pan F S, et al.A model for predicting grain sizes of As-cast Mg-Al-Ca alloys based on an artificial neural network with parameter optimization[J]. Mater. Rev., 2011, 25(9): 117(刘彬, 汤爱涛, 潘复生等. 基于参数优选的人工神经网络的Mg-Al-Ca系铸态合金晶粒尺寸预测模型[J]. 材料导报, 2011, 25(9): 117)
[8] Tang J L, Cai C Z, Xiao T T, et al.Application of support vector regression for Zr-2 alloy grain size prediction[J]. Trans. Mater. Heat Treat., 2013, 34(2): 180(唐江凌, 蔡从中, 肖婷婷等. 支持向量回归在Zr-2合金晶粒尺寸预测中的应用[J]. 材料热处理学报, 2013, 34(2): 180)
[9] Huang G B, Zhu Q Y, Siew C K.Extreme learning machine: Theory and applications[J]. Neurocomputing, 2006, 70: 489
[10] Han Y B.Gas emission prediction based on GA-ELM[J]. Saf. Coal. Min., 2015, 46(4): 166(韩义波. 基于GA-ELM的瓦斯涌出量预测[J]. 煤矿安全, 2015, 46(4): 166)
[11] Zhang X Y, Huang Q Q, Yin Z P, et al.Establishing a parametric flight loads identification method with GA-ELM model[J]. Adv. Aeronaut. Sci. Eng., 2014, 5: 497(张夏阳, 黄其青, 殷之平等. 基于GA-ELM的飞行载荷参数识别[J]. 航空工程进展, 2014, 5: 497)
[12] Wang X C, Shi F, Yu L, et al.43 Cases Analysis Based on MATLAB Neural Network [M]. Beijing: Beihang University Press, 2013: 243(王小川, 史峰, 郁磊等. MATLAB神经网络43个案例分析 [M]. 北京: 北京航空航天大学出版社, 2013: 243)
[13] Shamshirband S, Mohammadi K, Yee P L, et al.A comparative evaluation for identifying the suitability of extreme learning machine to predict horizontal global solar radiation[J]. Renew. Sustain. Energy Rev., 2015, 52: 1031
[14] Sajjadi S, Shamshirband S, Alizamir M, et al.Extreme learning machine for prediction of heat load in district heating systems[J]. Energy Build., 2016, 122: 222
[15] Wang J, Bi H Y.A new extreme learning machine optimized by PSO[J]. J. Zhengzhou Univ.(Nat. Sci. Ed.), 2013, 45(1): 100(王杰, 毕浩洋. 一种基于粒子群优化的极限学习机[J]. 郑州大学学报(理学版), 2013, 45(1): 100)
[16] Cao Z Y, Xia J C, Zhang M, et al.Optimization of gear blank preforms based on a new R-GPLVM model utilizing GA-ELM[J]. Knowl.-Based Syst., 2015, 83: 66
[17] Choudhury T A, Berndt C C, Man Z H.An extreme learning machine algorithm to predict the in-flight particle characteristics of an atmospheric plasma spray process[J]. Plasma Chem. Plasma Process., 2013, 33: 993
[18] Zhu W L.Application research of BP neural network based on genetic algorithm in multi-objective optimization [D]. Harbin: Harbin University of Science and Technology, 2009(朱文龙. 基于遗传算法的BP神经网络在多目标优化中的应用研究 [D]. 哈尔滨: 哈尔滨理工大学, 2009)
[19] Liu X L, Zhao X S, Lu F, et al.A GA-SVM based model for throwing rate prediction in the open-pit cast blasting[J]. J. China Coal Soc., 2012, 37: 1999(刘希亮, 赵学胜, 陆锋等. 基于GA-SVM的露天矿抛掷爆破抛掷率预测[J]. 煤炭学报, 2012, 37: 1999)
[20] Maity S P, Kundu M K.Genetic algorithms for optimality of data hiding in digital images[J]. Soft Comput., 2009, 13: 361
[21] Gu W, Li J Y, Wang Y D.Effect of grain size and Taylor factor on the transverse mechanical properties of 7050 aluminium alloy extrusion profile after over-aging[J]. Acta Metall. Sin., 2016, 52: 51(顾伟, 李静媛, 王一德. 晶粒尺寸及Taylor因子对过时效态7050铝合金挤压型材横向力学性能的影响[J]. 金属学报, 2016, 52: 51)
[22] Zhou J, Li X B, Shi X Z, et al.Predicting pillar stability for underground mine using Fisher discriminant analysis and SVM methods[J]. Trans. Nonferrous Met. Soc. China, 2011, 21: 2734
[23] Wu M W, Xiong S M.Microstructure simulation of high pressure die cast magnesium alloy based on modified CA method[J]. Acta Metall. Sin., 2010, 46: 1534(吴孟武, 熊守美. 基于改进CA方法的压铸镁合金微观组织模拟[J]. 金属学报, 2010, 46: 1534)
[24] Huang G B, Ding X J, Zhou H M.Optimization method based extreme learning machine for classification[J]. Neurocomputing2010, 74: 155
[25] Jiang X P, Dai Y J, Hong B.SVM model based on GA optimization for inertial prediction[J]. J. Nanjing Univ. Sci. Technol., 2011, 35(suppl.): 34(姜学鹏, 戴宇进, 洪贝. 基于遗传算法优化SVM模型的惯性器件故障预报[J]. 南京理工大学学报, 2011, 35(增刊): 34)
[26] Recker D, Franzke M, Hirt G, et al.Grain size prediction during open die forging processes[J]. Metall. Ital., 2010, 102(9): 29
[1] 刘震鹏, 闫志巧, 陈峰, 王顺成, 龙莹, 吴益雄. 金刚石工具用Cu-10Sn-xNi合金的制备和性能表征[J]. 金属学报, 2020, 56(5): 760-768.
[2] 赵燕春, 毛雪晶, 李文生, 孙浩, 李春玲, 赵鹏彪, 寇生中. Fe-15Mn-5Si-14Cr-0.2C非晶钢微观组织与腐蚀行为[J]. 金属学报, 2020, 56(5): 715-722.
[3] 梁孟超, 陈良, 赵国群. 人工时效对2A12铝板力学性能和强化相的影响[J]. 金属学报, 2020, 56(5): 736-744.
[4] 华涵钰,谢君,舒德龙,侯桂臣,盛乃成,于金江,崔传勇,孙晓峰,周亦胄. W含量对一种高W镍基高温合金显微组织的影响[J]. 金属学报, 2020, 56(2): 161-170.
[5] 王力,董超芳,张达威,孙晓光,Thee Chowwanonthapunya,满成,肖葵,李晓刚. 合金元素对铝合金在泰国曼谷地区初期腐蚀行为的影响[J]. 金属学报, 2020, 56(1): 119-128.
[6] 李鑫,董月成,淡振华,常辉,方志刚,郭艳华. 等通道角挤压制备超细晶纯Ti的腐蚀性能研究[J]. 金属学报, 2019, 55(8): 967-975.
[7] 马晋遥,王晋,赵云松,张剑,张跃飞,李吉学,张泽. 一种第二代镍基单晶高温合金1150 ℃原位拉伸断裂机制研究[J]. 金属学报, 2019, 55(8): 987-996.
[8] 吉华,邓运来,徐红勇,郭伟强,邓建峰,范世通. 焊接线能量对5182-O/HC260YD+Z异种材料CMT搭接接头组织与性能的影响[J]. 金属学报, 2019, 55(3): 376-388.
[9] 刘耀鸿,王朝辉,刘轲,李淑波,杜文博. Er对Mg-5Zn-xEr镁合金热裂敏感性的影响[J]. 金属学报, 2019, 55(3): 389-398.
[10] 邵成伟, 惠卫军, 张永健, 赵晓丽, 翁宇庆. 一种新型高强度高塑性冷轧中锰钢的组织和力学性能[J]. 金属学报, 2019, 55(2): 191-201.
[11] 田银宝, 申俊琦, 胡绳荪, 勾健. 丝材+电弧增材制造钛/铝异种金属反应层的研究[J]. 金属学报, 2019, 55(11): 1407-1416.
[12] 姚彦桃, 陈礼清, 王文广. 原位反应浸渗法制备(B4C+Ti)混杂增强Mg及AZ91D复合材料及其阻尼性能[J]. 金属学报, 2019, 55(1): 141-148.
[13] 赵晓丽, 张永健, 邵成伟, 惠卫军, 董瀚. 两相区退火处理冷轧0.1C-5Mn中锰钢的氢脆敏感性[J]. 金属学报, 2018, 54(7): 1031-1041.
[14] 翟斌, 周凯, 吕鹏, 王海鹏. 自由落体条件下Ti-6Al-4V合金微液滴的快速凝固研究[J]. 金属学报, 2018, 54(5): 824-830.
[15] 李金富, 周尧和. 液态金属深过冷快速凝固过程中初生固相的重熔[J]. 金属学报, 2018, 54(5): 627-636.