Ultra-high Precision Assembly of Aerospace Composite Structures: Fusing Physics-Based and Data-Driven Models

航空航天复合结构的超高精度组装：融合基于物理和数据驱动的模型

• 批准号: 2035038
• 负责人: Zhenyu Kong
• 金额: $ 31.94万
• 依托单位: Virginia Polytechnic Institute and State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-01-15 至 2024-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2035038&HistoricalAwards=false
• 关键词: Ultra; Precision; Assembly; Aerospace; Composite

Composite structures have increasingly emerged for aerospace and other applications because of their high strength-to-weight ratio, good resistance to harsh environments, and great performance reliability. Because composite structures have nonlinear, anisotropic, and compliant properties as well as inherent manufacturing variability, conventional process modeling and quality control methodologies for metal structure assembling are not adequate to composite components. This award supports fundamental research that integrates physics-guided and machine-learning models to advance ultra-high precision assembly of aerospace composite structures. The research involves seamless integration of physical and digital product connections, data science, advanced statistics, and comprehensive manufacturing knowledge. The research has the potential to minimize the material loss, decrease the production flow time and achieve high productivity and quality control for aerospace manufacturing. The scientific findings from this project may also be extended to other composites demanding industries, e.g., automotive, spacecraft and solar energy, and thus, increase the global competitiveness of the U.S. industry. The interdisciplinary nature of this project will provide students with unique educational and research experiences and cultivate a diverse and qualified workforce cognizant with combined manufacturing and data analytics abilities. In addition, the project will develop new learning modules for an undergraduate core course, recruit and mentor underrepresented students, offer industrial short courses, and develop open-source software for precision assembly, all potentially leading to profound impacts to the society.This research aims to develop fundamental knowledge and transformative technologies for ultra-high precision assembly of large complex-shaped composite structures by advancing physics-guided machine learning. Specific research activities include: (1) developing digital twin for ultra-high precision assembly of composite structures, (2) conducting physics-constrained active learning with safe exploration and efficient exploitation for experimental design and predictive modeling, (3) studying sparse machine learning for an optimal actuating strategy in composite structures assembly, and (4) analyzing theoretical properties of the methodologies. The project will research small-sized key aerospace structures as well as large-scale carbon fiber reinforced composite fuselage of ultra-high precision, e.g., less than 0.2 mm for a diameter of about 5 m. The methodologies will be computationally and experimentally evaluated in the verification and validation phase. The research outcomes will (1) expand the scientific understanding of engineering-driven data analytics and ultra-high precision quality control theory, (2) bridge the knowledge gap between predictive modeling, active learning, sparse learning, and composite structures assembly, and ultimately, (3) realize the effective all-inclusive integration of machine learning methodologies with advanced manufacturing.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.

复合材料结构因其高强度重量比、良好的耐恶劣环境性能和高可靠性而越来越多地出现在航空航天和其他应用中。由于复合材料结构具有非线性、各向异性和柔顺性以及固有的制造变异性，传统的金属结构装配过程建模和质量控制方法不适用于复合材料部件。该奖项支持整合物理指导和机器学习模型的基础研究，以推进航空复合材料结构的超高精度组装。该研究涉及物理和数字产品连接、数据科学、高级统计和综合制造知识的无缝集成。该研究具有降低材料损耗、缩短生产流程时间、实现高生产率和高质量控制的潜力。该项目的科学发现也可以扩展到其他复合材料要求行业，例如汽车，航天器和太阳能，从而提高美国工业的全球竞争力。该项目的跨学科性质将为学生提供独特的教育和研究经验，培养具有综合制造和数据分析能力的多样化和合格的劳动力。此外，该项目将为本科核心课程开发新的学习模块，招募和指导代表性不足的学生，提供工业短期课程，并开发用于精密装配的开源软件，所有这些都可能对社会产生深远的影响。本研究旨在通过推进物理引导的机器学习，为大型复杂形状复合材料结构的超高精度组装开发基础知识和变革性技术。具体的研究活动包括：(1)开发用于复合材料结构超高精度装配的数字孪生；(2)在实验设计和预测建模中进行物理约束下的主动学习安全探索和有效利用；(3)研究稀疏机器学习用于复合材料结构装配的最佳驱动策略；(4)分析方法的理论性质。该项目将研究小型关键航空结构，以及超大尺寸的超高精度碳纤维增强复合材料机身，例如直径约为5米，尺寸小于0.2毫米。这些方法将在验证和确认阶段进行计算和实验评估。研究成果将(1)扩展对工程驱动数据分析和超高精度质量控制理论的科学理解；(2)弥合预测建模、主动学习、稀疏学习和复合材料结构装配之间的知识鸿沟；(3)最终实现机器学习方法与先进制造的有效全面集成。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。

项目成果

MVGCN: Multi-View Graph Convolutional Neural Network for Surface Defect Identification Using Three-Dimensional Point Cloud

MVGCN：使用三维点云进行表面缺陷识别的多视图图卷积神经网络

• DOI: 10.1115/1.4056005
• 发表时间: 2023
• 期刊: Journal of Manufacturing Science and Engineering
• 作者: Wang, Yinan;Sun, Wenbo;Jin, Jionghua;Kong, Zhenyu;Yue, Xiaowei
• 通讯作者: Yue, Xiaowei

Failure-Averse Active Learning for Physics-Constrained Systems

• DOI: 10.1109/tase.2022.3213827
• 发表时间: 2021-10
• 期刊: IEEE Transactions on Automation Science and Engineering
• 影响因子: 5.6
• 作者: Cheolhei Lee;Xing Wang;Jianguo Wu;Xiaowei Yue

Nested Bayesian Optimization for Computer Experiments

• DOI: 10.1109/tmech.2022.3202079
• 发表时间: 2021-12
• 期刊: IEEE/ASME Transactions on Mechatronics
• 作者: Yan Wang;M. Wang;Areej AlBahar;Xiaowei Yue

Gaussian Processes with Input Location Error and Applications to the Composite Parts Assembly Process

• DOI: 10.1137/20m1312447
• 发表时间: 2020-02
• 期刊: SIAM/ASA J. Uncertain. Quantification
• 作者: Wenjia Wang;Xiaowei Yue;Ben Haaland;C. F. Wu

SPAC: S parse sensor p lacement-based a daptive c ontrol for high precision fuselage assembly

SPAC：基于 Sparse 传感器放置的自适应控制，用于高精度机身组装

• DOI: 10.1080/24725854.2022.2116133
• 发表时间: 2022
• 期刊: IISE Transactions
• 影响因子: 2.6
• 作者: Mou, Shancong;Biehler, Michael;Yue, Xiaowei;Hunt, Jeffrey H.;Shi, Jianjun
• 通讯作者: Shi, Jianjun