DMREF/GOALI/Collaborative Research: Physics-Informed Artificial Intelligence for Parallel Design of Metal Matrix Composites and their Additive Manufacturing

这项旨在革新和改造我们未来的设计材料（DMREF）研究使物理知情的人工智能（AI）能够设计用陶瓷颗粒（金属基复合材料）增强的金属材料及其增材制造（3D打印）。相对于没有陶瓷增强的相同金属材料，这种材料可以在更高的温度下表现出优越的机械性能。增材制造为由金属基复合材料制成的高性能、轻质结构部件提供了前所未有的制造能力。然而，金属基复合材料的设计及其增材制造在很大程度上是通过昂贵、耗时的试验和错误方法进行的；这类零件的质量保证同样受到挑战。人工智能指导的材料设计和鉴定及其制造可以显着降低此类技术的时间和成本障碍。在该计划中进行的基础研究将填补关键空白，使人工智能能够发现和优化这些材料及其制造，从而将部署时间和成本减少一半，以满足材料基因组计划的愿景。外展项目和多样性、公平性和包容性计划包括人工智能制造课程，涵盖幼儿园至毕业生，其中包括从该项目开发的示例问题和工具。亚特兰大和盐湖城的高中教师和来自少数族裔的学生将在这些课程中获得实践经验和指导。该研究维持并扩展了支持合金、陶瓷及其复合材料基础研究的强大项目；支持大学（佐治亚理工学院和犹他州）、初创企业（GOALI合作伙伴Elementum 3D）和国家实验室（空军研究实验室）之间自由流动互动的模式；扩大对自动化材料制造研究的投资，确保到2030年美国在该领域处于领先地位；所有这些都在适当的时候使用计算方法、数据分析、机器学习和自主实验3D表征。该研究计划使物理知情的人工智能（AI）驱动的金属基复合材料及其增材制造并行设计成为可能。同时发现和优化新材料及其增材制造（AM）的人工智能概念有望进一步革新AM，但尚未实现。基础研究是为了实现材料及其制造的自主人工智能发现和优化，从而将部署时间和成本减少一半，以满足材料基因组计划的愿景。五个关键的数据驱动算法缺口将被填补：1)数据分析-解释-管理算法，以实现从必要的过程-结构-属性数据源自动的、谱系的数据管理。2)自动清理数据和连接数据库的算法，当新的数据源或数据特征被纳入研究问题时，人工智能可以修改和附加数据空间。3)自动化跨多个长度和时间尺度的数据特征映射算法，以完成过程-结构-属性数据本体。4)提高人工智能性能的数据特征工程算法。5)跨多个嵌套子模型学习全局关系的过程-结构-属性机器学习模型。将推进基于物理的模型和实验，以预测和验证它们在发现和优化金属基复合材料及其在多个长度和时间尺度上的增材制造中的效用。高通量的一维、二维和三维特征数据分析将实现自动化。GOALI合作伙伴Elementum 3D将提供一项用于增材制造的新型金属基复合材料商业化的技术经济基线研究，作为该计划取得进展的总体评估指标。新的AM测试工件的开发将使全球的研究人员受益。为自动化数据工作流程开发的协议和标准可以通过增加对各种材料和制造工艺的高通量和高保真数据源的访问，使世界各地的材料科学和工程研究人员受益，包括机器学习模型和人工智能知识系统。该项目由工程理事会（ENG）的土木、机械和制造创新司（CMMI）、计算机和信息科学与工程理事会（CISE）的信息和智能系统司（IIS）以及数学和物理科学理事会（MPS）的材料研究司（DMR）共同资助。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This Designing Materials to Revolutionize and Engineer our Future (DMREF) research enables physics-informed artificial intelligence (AI) design of metal materials reinforced with ceramic particles (metal matrix composites) and their additive manufacturing (3D printing). Such materials can exhibit superior mechanical performances at higher temperatures relative to the same metal material without ceramic reinforcements. Additive manufacturing provides unprecedented fabrication capability for high performance, lightweight structural components made from metal matrix composite materials. However, the design of metal matrix composites and their additive manufacturing is largely performed with expensive, time consuming trial and error methodologies; quality assurance of such parts is similarly challenged. AI-guided design and qualification of materials and their manufacturing can significantly lower the time and cost barriers to such technologies. The basic research performed in this program will fill critical gaps to enable AI discovery and optimization of these materials and their manufacturing toward reducing deployment times and costs by half, to meet the Materials Genome Initiative vision. The outreach programs and diversity, equity, and inclusion plans include AI manufacturing course curricula spanning kindergarten - graduate which include example problems and tools developed from this program. Atlanta and Salt Lake City high school teachers and students from underrepresented minority populations will receive hands-on experience and instruction in these curricula. The research maintains and expands robust programs supporting fundamental research in alloys, ceramics, and their composites; support modalities for free-flowing interactions among universities (Georgia Tech and Utah), start-up ventures (GOALI partner Elementum 3D), and national laboratories (Air Force Research Laboratory); expand investments in automated materials manufacturing research to ensure the U.S. is the leader in the field by 2030; all using, when appropriate, computational methods, data analytics, machine learning, and autonomous experimental 3D characterization.This research program enables physics-informed artificial intelligence (AI) - driven parallel design of metal matrix composites and their additive manufacturing. The concept of AI that discovers and optimizes new materials and their Additive Manufacturing (AM) in parallel promises to further revolutionize AM but is yet to be realized. Basic research is to enable autonomous AI discovery and optimization of materials and their manufacturing toward reducing deployment times and costs by half, to meet the Materials Genome Initiative vision. Five critical data-driven algorithmic gaps will be filled: 1) data analysis-interpretation-curation algorithms to enable automatic, pedigreed data curation from requisite process-structure-property data sources. 2) Algorithms that automate data cleaning and concatenation of databases so that AI can modify and append the data spaces when new data sources or data features are incorporated into a research problem. 3) Algorithms that automate data feature mapping across multiple length and time scales to complete process-structure-property data ontologies. 4) Data feature engineering algorithms that improve the AI performance. 5) Process-structure-property machine learning models that learn global relationships across multiple nested submodels. Physics-based models and experiments will be advanced to predict and verify their utility in discovering and optimizing metal matrix composites and their additive manufacturing at multiple length and time scales. High throughput one-dimensional, two-dimensional, and three-dimensional characterization data analyses will be automated. GOALI partner Elementum 3D will provide a techno-economic baseline study of commercializing a new metal matrix composite for additive manufacturing to be used as an overall assessment metric for the advancements made in this program. The development of a new AM test artifact will benefit researchers around the globe. The protocols and standards developed for automating data workflows can benefit materials science and engineering researchers around the world by increasing access to high-throughput and high-fidelity data sources, including machine learning models and AI knowledge systems, for all kinds of materials and manufacturing processes.This project is jointly funded by the Division of Civil, Mechanical and Manufacturing Innovation (CMMI) in the Directorate for Engineering (ENG), the Division of Information and Intelligent Systems (IIS) in the Directorate for Computer and Information Science and Engineering (CISE), and the Division of Materials Research (DMR) in the Directorate for Mathematical and Physical Sciences (MPS).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.