机器学习在金属增材制造中的应用现状和前景展望  

Current Situation and Future Prospect of Machine Learning in Metal Additive Manufacturing

作　　者：刘志远[1] 丁卯 王沛 陈张伟 杨灿 徐斌[1] 彭太江[1] 刘长勇[1] 沈军 LIU Zhiyuan;DING Mao;WANG Pei;CHEN Zhangwei;YANG Can;XU Bin;PENG Taijiang;LIU Changyong;SHEN Jun(Shenzhen University,Shenzhen 518060,China;Shenzhen Technology University,Shenzhen 518118,China)

机构地区：[1]深圳大学,深圳518060 [2]深圳技术大学,深圳518118

出　　处：《航空制造技术》2022年第23期14-28,共15页Aeronautical Manufacturing Technology

基　　金：国家自然科学基金(51971149)。

摘　　要：增材制造作为一种先进的智能制造技术,能够直接快速、方便高效地制备各种复杂结构金属零部件,受到广泛关注,目前已应用于航空航天、医疗器械等先进制造领域。然而,由于适用于增材制造的金属材料成分有限以及复杂的打印过程容易产生缺陷等问题,使得该技术的大范围工业应用受到了一定程度的限制。机器学习具有出色的数据处理和分析能力,被广泛应用于日常生活和工业生产各领域以提高智能化水平。以金属材料增材制造的工艺窗口建立、打印质量控制、打印金属内部微观组织结构分析、力学性能探究为主线,综述了机器学习在增材制造过程中的应用,讨论了机器学习在金属增材制造领域的发展机遇和挑战,展望了该领域未来的发展前景。As an advanced intelligent manufacturing technology,additive manufacturing(AM)can directly produce metallic components with complex macroscopic structure in short lead time and has attracted lots of attention in recent years.It has been widely used in many advanced manufacture fields such as aerospace,medical device,and so on.However,there exhibit limited number of alloy systems suitable for AM printing,and the complex printing process makes it easy to introduce defects.Consequently,the large-scale application of AM is hindered.Machine learning has been widely used in various daily life and industrial production fields due to its excellent data processing and analysis capabilities.In this paper,the application of machine learning in the AM process including processing window establishing,printing quality control,printed metallic microstructure identification,and mechanical properties exploration are reviewed.In the end,the opportunities and challenges of machine learning in AM are discussed,and the further research directions are proposed.

关 键 词：增材制造(AM) 机器学习 工艺参数优化 缺陷检测 组织结构调控 力学性能预测 

分 类 号：TG14[一般工业技术—材料科学与工程] TP391.73[金属学及工艺—金属材料] TP181[金属学及工艺—金属学]