结构驱动的物理信息深度学习用于有限数据的激光粉末床聚变增材制造中的温度预测

IF 2.4 3区 工程技术 Q3 ENGINEERING, MANUFACTURING
Suyog Ghungrad, Meysam Faegh, B. Gould, S. Wolff, Azadeh Haghighi
{"title":"结构驱动的物理信息深度学习用于有限数据的激光粉末床聚变增材制造中的温度预测","authors":"Suyog Ghungrad, Meysam Faegh, B. Gould, S. Wolff, Azadeh Haghighi","doi":"10.1115/1.4062237","DOIUrl":null,"url":null,"abstract":"\n Physics-informed deep learning (PIDL) is one of the emerging topics in additive manufacturing (AM). However, the success of previous PIDL approaches is generally significantly dependent on the existence of massive datasets. As the data collection in AM is usually challenging, a novel Architecture-driven PIDL structure named APIDL based on the deep unfolding approach for limited data scenarios has been proposed in the current study for predicting thermal history in the laser powder bed fusion process. The connections in this machine learning architecture are inspired by iterative thermal model equations. In other words, each iteration of the thermal model is mapped to a layer of the neural network. The hyper-parameters of the APIDL model are tuned, and its performance is analyzed. The APIDL for 1000 points with 80:20 split ratio achieves a testing mean absolute percentage error (MAPE) of 2.8% and R2 value of 0.936. The APIDL is compared with the artificial neural network, extra trees regressor (ETR), support vector regressor, and long short-term memory algorithms. It was shown that the proposed APIDL model outperforms the others. The MAPE and R2 of APIDL are 55.7% lower and 15.6% higher than the ETR, which had the best performance among other pure ML models.","PeriodicalId":16299,"journal":{"name":"Journal of Manufacturing Science and Engineering-transactions of The Asme","volume":null,"pages":null},"PeriodicalIF":2.4000,"publicationDate":"2023-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Architecture-driven Physics-informed Deep Learning for Temperature Prediction in Laser Powder Bed Fusion Additive Manufacturing with Limited Data\",\"authors\":\"Suyog Ghungrad, Meysam Faegh, B. Gould, S. Wolff, Azadeh Haghighi\",\"doi\":\"10.1115/1.4062237\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"\\n Physics-informed deep learning (PIDL) is one of the emerging topics in additive manufacturing (AM). However, the success of previous PIDL approaches is generally significantly dependent on the existence of massive datasets. As the data collection in AM is usually challenging, a novel Architecture-driven PIDL structure named APIDL based on the deep unfolding approach for limited data scenarios has been proposed in the current study for predicting thermal history in the laser powder bed fusion process. The connections in this machine learning architecture are inspired by iterative thermal model equations. In other words, each iteration of the thermal model is mapped to a layer of the neural network. The hyper-parameters of the APIDL model are tuned, and its performance is analyzed. The APIDL for 1000 points with 80:20 split ratio achieves a testing mean absolute percentage error (MAPE) of 2.8% and R2 value of 0.936. The APIDL is compared with the artificial neural network, extra trees regressor (ETR), support vector regressor, and long short-term memory algorithms. It was shown that the proposed APIDL model outperforms the others. The MAPE and R2 of APIDL are 55.7% lower and 15.6% higher than the ETR, which had the best performance among other pure ML models.\",\"PeriodicalId\":16299,\"journal\":{\"name\":\"Journal of Manufacturing Science and Engineering-transactions of The Asme\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":2.4000,\"publicationDate\":\"2023-03-30\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Manufacturing Science and Engineering-transactions of The Asme\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://doi.org/10.1115/1.4062237\",\"RegionNum\":3,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"ENGINEERING, MANUFACTURING\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Manufacturing Science and Engineering-transactions of The Asme","FirstCategoryId":"5","ListUrlMain":"https://doi.org/10.1115/1.4062237","RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"ENGINEERING, MANUFACTURING","Score":null,"Total":0}
引用次数: 0

摘要

物理知情深度学习(PIDL)是增材制造领域的新兴课题之一。然而,以前PIDL方法的成功通常在很大程度上取决于海量数据集的存在。由于AM中的数据收集通常具有挑战性,在当前的研究中,基于有限数据场景的深度展开方法,提出了一种新的架构驱动的PIDL结构APIDL,用于预测激光粉末床聚变过程中的热历史。这种机器学习架构中的连接受到迭代热模型方程的启发。换句话说,热模型的每次迭代都被映射到神经网络的一层。对APIDL模型的超参数进行了调整,并对其性能进行了分析。对于具有80:20分割比的1000个点,APIDL实现了2.8%的测试平均绝对百分比误差(MAPE)和0.936的R2值。将APIDL与人工神经网络、额外树回归器(ETR)、支持向量回归器和长短期记忆算法进行了比较。结果表明,所提出的APIDL模型优于其他模型。APIDL的MAPE和R2分别比ETR低55.7%和15.6%,在其他纯ML模型中性能最好。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Architecture-driven Physics-informed Deep Learning for Temperature Prediction in Laser Powder Bed Fusion Additive Manufacturing with Limited Data
Physics-informed deep learning (PIDL) is one of the emerging topics in additive manufacturing (AM). However, the success of previous PIDL approaches is generally significantly dependent on the existence of massive datasets. As the data collection in AM is usually challenging, a novel Architecture-driven PIDL structure named APIDL based on the deep unfolding approach for limited data scenarios has been proposed in the current study for predicting thermal history in the laser powder bed fusion process. The connections in this machine learning architecture are inspired by iterative thermal model equations. In other words, each iteration of the thermal model is mapped to a layer of the neural network. The hyper-parameters of the APIDL model are tuned, and its performance is analyzed. The APIDL for 1000 points with 80:20 split ratio achieves a testing mean absolute percentage error (MAPE) of 2.8% and R2 value of 0.936. The APIDL is compared with the artificial neural network, extra trees regressor (ETR), support vector regressor, and long short-term memory algorithms. It was shown that the proposed APIDL model outperforms the others. The MAPE and R2 of APIDL are 55.7% lower and 15.6% higher than the ETR, which had the best performance among other pure ML models.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
CiteScore
6.80
自引率
20.00%
发文量
126
审稿时长
12 months
期刊介绍: Areas of interest including, but not limited to: Additive manufacturing; Advanced materials and processing; Assembly; Biomedical manufacturing; Bulk deformation processes (e.g., extrusion, forging, wire drawing, etc.); CAD/CAM/CAE; Computer-integrated manufacturing; Control and automation; Cyber-physical systems in manufacturing; Data science-enhanced manufacturing; Design for manufacturing; Electrical and electrochemical machining; Grinding and abrasive processes; Injection molding and other polymer fabrication processes; Inspection and quality control; Laser processes; Machine tool dynamics; Machining processes; Materials handling; Metrology; Micro- and nano-machining and processing; Modeling and simulation; Nontraditional manufacturing processes; Plant engineering and maintenance; Powder processing; Precision and ultra-precision machining; Process engineering; Process planning; Production systems optimization; Rapid prototyping and solid freeform fabrication; Robotics and flexible tooling; Sensing, monitoring, and diagnostics; Sheet and tube metal forming; Sustainable manufacturing; Tribology in manufacturing; Welding and joining
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信