FMSG: Cyber: Using a cloud-based platform to quantify the uncertainty of the process-structure-property-surface relationship for repeatable additive manufacturing of Inconel 718

该未来制造种子基金(FMSG)项目支持对使用金属添加剂制造(AM)作为功能性终端部件的生产过程所需的理论和实验基础的研究。金属AM是美国制造业创新和竞争力的驱动力，因为它允许快速实施新的部件设计，以缩短新产品的上市时间。然而，使用金属AM作为生产过程而不是原型工具，需要可靠和重复地制造具有几乎相同的结构、表面形貌和性能的部件。因此，了解与工艺-结构-性能-表面(PSPS)关系相关的不确定性是重要的。然而，PSPS研究是耗时和昂贵的，因为需要许多样本来得出有意义的信息，或者，由于访问/许可和数据格式之间的不一致，汇总不同研究的现有数据集以扩大和加强对PSPS关系的洞察并不简单。因此，本研究旨在通过利用不确定性量化(UQ)框架和机器学习(ML)算法来分析微结构和表面形貌图像，并量化PSPS关系，从而解决这些基本问题。此外，基于云的数据库将使数据和知识可供其他研究人员使用。通过与弗吉尼亚理工大学(弗吉尼亚理工大学)、弗吉尼亚州立大学(弗吉尼亚州立大学)、历史上四年的黑人学院/大学(HBCU)和弗吉尼亚州公私合作伙伴关系联邦先进制造中心(CCAM)的合作，研究与教育和劳动力发展(特别是代表不足的群体)相集成。该项目的研究目标有两个：量化因科奈尔718的激光粉床融合(L-PBF)PSPS关系的不确定性，并建立一个基于云的数据库，以在不同研究人员之间聚合和共享PSPS数据，以减少重复工作，并通过数据共享加快PSPS研究。为了实现这一目标，本研究旨在将UQ框架与ML算法相结合，以得出数据驱动模型，这些模型将L-PBF工艺参数与量化微观结构和竣工表面形貌的指标联系起来。从这项研究中获得的知识将(1)量化作为L-PBF工艺参数(正向UQ问题)的函数的显微组织和竣工表面形貌的不确定性；(2)确定获得微观结构和竣工表面形貌的特定不确定性(或概率定义)所需的L-PBF工艺参数(UQ逆问题)；(3)推导出正问题和逆问题的解及其作为L-PBF工艺参数的函数的不确定性的操作图；(4)实现基于云的数据库，以聚合可以使用数字对象标识符来引用的微结构图像和表面地形图，并能够组合用户生成的数据集。该项目的成果将减少技术壁垒，并鼓励采用金属AM作为功能终端部件的可行制造工艺。这项未来的制造研究由计算机和信息科学与工程局的计算机和网络系统部门(CESE/CNS)和社会、行为和经济科学局的社会科学和经济学部门(SBE/SES)支持。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This Future Manufacturing Seed Grant (FMSG) project supports research into the theoretical and experimental foundation required to use metal additive manufacturing (AM) as a production process for functional end-use parts. Metal AM is a driver for innovation and competitiveness of United States manufacturing because it allows rapid implementation of new part designs to reduce the time-to-market of new products. However, using metal AM as a production process instead of a prototyping tool requires reliably and repeatably manufacturing parts with near-identical structure, surface topography, and properties. Consequently, understanding the uncertainty associated with the process-structure-property-surface (PSPS) relationship is important. Yet, PSPS research is time-consuming and costly because many specimens are required to derive meaningful information and, alternatively, aggregating existing datasets of different studies to expand and enhance insights about the PSPS relationship is not straightforward because of access/permissions and inconsistencies between data formats. Hence, this research specifically aims to address these fundamental problems by leveraging an uncertainty quantification (UQ) framework and machine learning (ML) algorithms to analyze microstructure and surface topography images, and quantify the PSPS relationship. Additionally, a cloud-based database will make the data and knowledge available to other researchers. Research integrates with education and workforce development, specifically underrepresented groups, through a partnership with Virginia Tech (VT), Virginia State University (VSU), a 4-year Historically Black College/University (HBCU), and the Commonwealth Center for Advanced Manufacturing (CCAM), a public-private partnership in Virginia.The research objective of this project is twofold: quantify the uncertainty of the PSPS relationship for laser powder bed fusion (L-PBF) of Inconel 718 and, establish a cloud-based database to aggregate and share PSPS data among different researchers, to reduce duplication of effort and accelerate PSPS research by data-sharing. To accomplish this objective, the research aims to combine a UQ framework with ML algorithms to derive data-driven models that relate L-PBF process parameters to metrics that quantify the microstructure and as-built surface topography. The knowledge resulting from this research will (1) quantify the uncertainty of the microstructure and as-built surface topography as a function of the L-PBF process parameters (forward UQ problem); (2) determine the L-PBF process parameters required to obtain specific uncertainty (or probability definition) of microstructure and as-built surface topography (inverse UQ problem); (3) derive an operating map of the solution of the forward and inverse problems and its uncertainty as a function of the L-PBF process parameters; (4) implement a cloud-based database to aggregate microstructure images and surface topography maps that can be cited using a digital object identifier, and enable combining user-generated datasets. The outcomes of this project will reduce technical barriers and spur adoption of metal AM as a viable manufacturing process for functional end-use parts.This Future Manufacturing research is supported by the Computer and Information Science and Engineering Directorate's Division of Computer and Network Systems (CISE/CNS) and the Social, Behavioral and Economic Sciences Directorate’s Division of Social and Economic Sciences (SBE/SES).This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.