EAGER/Collaborative Research: Explore the Theoretical Framework of Engineering Knowledge Transfer in Cybermanufacturing Systems

3D打印，也被称为增材制造，为个人和公司提供了设计和生产定制零件的潜力，批量小到单个单元。然而，与传统的大规模生产不同，生产机器通过定期测量所生产的零件来调整以保持质量，3D打印生产的小批量不提供以同样的方式调整生产机器的机会。这项探索性研究（EAGER）的早期概念资助将研究一种统计方法，根据生产不同零件类型的零件的经验来调整3D打印机的生产质量。这种工具能够作为基于云的实用程序提供给制造商，并可能扩展到聚合来自不同站点的类似机器的数据。工程过程之间知识转移的根本障碍在于潜伏变量，潜伏变量是由于测量不可行或知识不足而无法观察到的过程变量。该项目将根据可观察的控制变量建模并减轻潜伏变量的影响，从而为增材制造系统实现一种新的灰盒模型转移策略。提出的研究任务包括：(1)建立效应等效的理论公式以量化潜伏变量的效应；(2)探索学习效应等效的统计基础；(3)验证效应等效模型及其稳健性；(4)从效应等效中获得工程洞察力并制定多分辨率测量策略。该项目预计将产生工程效果等效的数学公式，证明该公式在预测增材制造机器制造的3D产品的几何形状变形方面的作用，以及基于等效的增材制造过程控制的多分辨率测量策略。

3D Printing, also known as additive manufacturing, offers the potential for individuals and companies to design and produce customized parts in quantities as small as a single unit. However, unlike conventional mass production, for which production machines are adjusted to maintain quality by periodically measuring the parts that are produced, the small lots produced in 3D Printing do not provide the opportunity to tune the production machines in the same way. This EArly-concept Grant for Exploratory Research (EAGER) will investigate a statistical method for adjusting the production quality of a 3D Printing machine based on experience in producing parts of different part types. Such a tool has the capability to be offered to manufacturers as a cloud-based utility, with possible extension to aggregating data from similar machines at different sites.The fundamental barrier to knowledge transfer between engineering processes lies in lurking variables, which are process variables that are unobserved due to infeasibility of measurement or insufficient knowledge. The project will model and mitigate the effects of lurking variables in terms of observable control variables to enable a novel gray-box model transfer strategy for additive manufacturing systems. The proposed research tasks include: (1) establishing a theoretical formulation of effect equivalence to quantify effects of lurking variables, (2) exploring a statistical foundation for learning effect equivalence, (3) verifying effect equivalence models and their robustness, and (4) obtaining engineering insight from effect equivalence and developing a multi-resolution measurement strategy. The project is expected to produce a mathematical formulation of engineering effect equivalence, demonstrate the formulation in predicting geometric shape deformation of 3D products built by additive manufacturing machines, and a multi-resolution measurement strategy for equivalence-based additive manufacturing process control.