EAGER/Cybermanufacturing Systems: Fleet-Sourced Cyber Manufacturing Applications for Improved Transparency and Resilience of Manufacturing Assets and Systems

支持互联网的服务（如基于云的和移动应用程序）在零售、音乐、交通和医疗保健等几乎所有经济部门都具有影响力，这些部门已经证明了对来自网络系统的历史数据进行分析的好处。然而，与现有的互联网支持行业相比，制造业资产的连接程度较低，并且无法实时访问。因此，目前的制造企业的决策遵循自上而下的方法：从整体设备效率到生产要求的分配，而不考虑机器的状况。这项早期概念探索性研究（EAGER）资助支持基础研究，以开发下一代先进网络制造系统的概念和理论，这些系统通过对舰队来源数据的分析实现网络化和互操作。通过考虑基于历史资产数据预测的资产健康状况，网络制造系统将实现自下而上的实时决策支持制造策略。最终，移动应用程序将被开发用于可操作信息的便携式访问。由于这种分析可以对从现有资产状态监测系统收集的数据进行，并且可以安装适度水平的附加传感器，因此几乎每个行业的制造业都将从这项研究的结果中受益。因此，本研究将通过提高制造业企业的效率和生产率，为美国经济的发展注入速度，造福社会。这项研究需要来自不同学科的知识和专业知识，包括制造、机械工程、计算机科学和控制理论。跨学科的方法将促进相关领域的创造力和健康发展，并吸引年轻一代的兴趣，以影响科学，技术，工程和数学教育。新的网络制造方法将深化对机队来源预测的研究，它克服了传统预测和健康管理方法的几个缺点，包括缺乏通用性和可重构性，缺乏对变化制度的鲁棒性，有时准确性不足。车队是指在工作条件（品牌和型号、环境条件和健康状态）方面相似的一组资产。船队源预测的研究空白存在于资产相似度的量化、船队聚类、船队验证以及制度变化时动态改变聚类方案等方面。研究团队将利用现有的车队级点对点预测方法来开发一个可重构平台，该平台具有从车队来源的数据中降低维度的能力，设计一种风险评估方法来提供实时预测可操作信息，并最终将这些功能整合到开发的移动应用程序中。

Internet-enabled services (such as cloud-based and mobile applications) have been influential through almost all economic sectors, such as retail, music, transportation, and healthcare, which have proven the benefit of performing analytics on historical data from a networked system. However, compared to existing Internet-enabled industries, manufacturing assets are less connected and less accessible in real-time. As a result, current manufacturing enterprises make decisions following a top-down approach: from overall equipment effectiveness to assignment of production requirements, without considering the condition of machines. This EArly-concept Grant for Exploratory Research (EAGER) award supports fundamental research to develop the concepts and theory for next-generation advanced cybermanufacturing systems that are networked and interoperable through analytics on the fleet-sourced data. Cybermanufacturing systems will enable a bottom-up real-time decision support manufacturing strategy by taking into account asset health conditions predicted based on historical asset data. Eventually, mobile applications will be developed for portable access of the actionable information. Since such analytics can be performed on data collected from existing asset condition monitoring systems with moderate levels of add-on sensor installment, manufacturing industries in almost every sector will benefit from the results of this research. Consequently, this research will inject speed into the development of U.S. economy and benefit the society by increasing the efficiency and productivity of manufacturing enterprises. This research requires knowledge and expertise from a variety of disciplines including manufacturing, mechanical engineering, computer science, and control theory. The interdisciplinary methodology will facilitate the creativity and healthy growth of the involved areas and draw interest from younger generation to impact science, technology, engineering and mathematics education.The new cybermanufacturing methodology will deepen the research on fleet-sourced prognostics, which overcomes several drawbacks of conventional prognostics and health management approaches, including lack of generality and reconfigurability, lack of robustness against changing regimes, and sometimes insufficient accuracy. A fleet is referred to as a group of assets similar in working conditions (make and model, ambient conditions, and health status). Research gaps in conducting fleet-sourced prognostics exist in the quantification of asset similarity, clustering fleets, validation of such fleets, and dynamically changing the clustering scheme when regimes change. The research team will leverage existing fleet-level peer-to-peer prognostics approaches to develop a reconfigurable platform with capabilities to reduce the dimensionality from fleet-sourced data, devise a risk assessment methodology to provide real-time predictive actionable information, and eventually incorporate such functions into the developed mobile applications.