激光熔化沉积Al-Mg-Sc-Zr合金单熔道成形质量预测  

Prediction of single-track forming quality in laser melting deposition of Al-Mg-Sc-Zr alloy

作　　者：肖罡 李伟奇[1] 谢莉 路超 刘筱 朱必武[1,4] 杨钦文 XIAO Gang;LI Weiqi;XIE Li;LU Chao;LIU Xiao;ZHU Biwu;YANG Qinwen(College of Mechanical and Electrical Engineering,Hunan University of Science and Technology,Xiangtan 411201,China;Jiangxi Copper Technology Research Institute Co.,Ltd.,Nanchang 330096,China;College of Mechanical and Vehicle Engineering,Hunan University,Changsha 410082,China;College of Materials Science and Engineering,Guangdong Ocean University,Yangjiang 529559,China)

机构地区：[1]湖南科技大学机电工程学院,湘潭411201 [2]江西铜业技术研究院有限公司,南昌330096 [3]湖南大学机械与运载工程学院,长沙410082 [4]广东海洋大学材料科学与工程学院,阳江529559

出　　处：《中国有色金属学报》2024年第4期1010-1021,共12页The Chinese Journal of Nonferrous Metals

基　　金：国家自然科学基金资助项目(52075159,52071139);江西省高层次高技能领军人才培养工程资助项目;江西省杰出青年基金资助项目(20224ACB218002);湖南省自然科学基金资助项目(2023JJ30252,2023JJ30262,2022JJ30019);湖南省教育厅科研项目(21B0471)。

摘　　要：本研究探讨了激光熔化沉积(LMD)工艺参数对Al-Mg-Sc-Zr合金单熔道成形质量的影响,并采用BP神经网络模型和改进的PSO-BP神经网络模型预测了不同工艺参数下的成形孔隙率。结果表明:Al-Mg-Sc-Zr合金的LMD单熔道成形孔隙率,随着激光线能量密度的增加呈现先上升后下降再上升的趋势,在175 J/mm^(2)时孔隙率达到最小值,仅为1.38%。在相同的送粉工艺参数下,LMD成形熔道宽度和熔道深度均与激光功率呈现出正相关性,而熔道高度和熔道深度均与扫描速率呈现出负相关性。相较于BP神经网络模型,改进的PSO-BP神经网络模型的平均绝对误差降低了29.69%,平均相对误差降低了19.42%,均方误差降低了19.02%,相关系数提升了63.44%。The influence of laser melting deposition(LMD)process parameters on the forming quality of a singletrack Al-Mg-Sc-Zr alloy was investigated.A BP neural network model and an improved PSO-BP neural network model were employed to predict the porosity rate of the formed track under different LMD parameters.The results demonstrate that the porosity rate firstly increases,then decreases,and finally increases with laser line energy density increasing.When the laser line energy density was 175 J/mm^(2),the porosity rate displays a minimum value,which is 1.38%.Under the same powder feeding rate,the width and depth of the track show a positive correlation with laser power,while the height and depth of the track illustrate a negative correlation with scanning speed.In contrast to the BP neural network model,the average absolute error of the improved PSO-BP neural network model is reduced by 29.69%,the average relative error of the improved PSO-BP neural network model is reduced by 19.42%,the mean square error of the improved PSO-BP neural network model is reduced by 19.02%and the correlation coefficient of the improved PSO-BP neural network model is improved by 63.44%.

关 键 词：激光熔化沉积 AL-MG-SC-ZR合金 熔池质量 孔隙率 神经网络 

分 类 号：TG665[金属学及工艺—金属切削加工及机床]