EAGER:REAL-D: Smart Decision Making using Data and Advanced Modeling Approaches

拟议的探索性研究项目旨在开发下一代自主制药生产过程，将产品和过程信息学与知识管理相结合。过程数据、过程模型和信息管理工具的集成将使操作条件的自适应调整成为可能，以补偿原材料的可变性和不断变化的产品需求。该研究团队将利用罗格斯大学结构化有机颗粒系统中心（C-SOPS）的设施进行原理验证研究，并生成实验数据，以促进对每个过程的基本理解。为了实现整个制造供应链向更加自主和分散的决策过渡，必须开发一个集成平台，以：（a）使用数据历史平台从所述制造设施获取关于过程和产品操作的数据;（B）利用所述数据来提取关于过程理解的进一步知识;以及（c）使用所述知识来动态地和自适应地改进过程操作。对于任务（a），建议使用数据管理系统，例如OSI PI，其具有从多个来源接收数据的能力，包括控制平台以及过程分析技术（PAT）数据管理工具。该平台能够使用事件框架功能建立配方层次结构，并定期将数据推送到云系统中，以实现永久的企业范围数据存储和高效共享。对于任务（B），建议使用先进的统计和机器学习方法，并结合数据核对方法。 最后，对于任务（c），将利用所获得的信息，通过建立准确的代理模型来调整模型可行空间，并使用在线数据采集来自适应地完善它们。虽然重点将放在制药生产过程上，但拟议的工作如果成功，将对各种工业过程产生更广泛的影响。该奖项反映了NSF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。

The proposed exploratory research project aims to develop a next-generation autonomous manufacturing process for pharmaceutical production that integrates product and process informatics with knowledge management. The integration of process data, process models, and information management tools will enable adaptive adjustment to the operating conditions to compensate for variability in raw materials and changing product needs. The research team will take advantage of the facilities of the Center for Structured Organic Particulate Systems (C-SOPS) at Rutgers University for proof of principle studies and generation of experimental data for advancing fundamental understanding of each process.To enable the transition towards more autonomous and de-centralized decisions across the entire manufacturing supply chain, it is imperative to develop an integrated platform to: (a) acquire data regarding process and product operations from the manufacturing facility using data historian platforms; (b) utilize the data to extract further knowledge on process understanding; and (c) use this knowledge to dynamically and adaptively improve process operations. For task (a), the use of a data management system, such as OSI PI, is proposed with the ability to receive data from multiple sources including the control platform as well as the Process Analytical Technology (PAT) data management tool. This platform has the capability to build up recipe hierarchical structure using Event Frame functionality and periodically push the data into a cloud system for permanent enterprise-wide data storage and efficient sharing. For task (b), the use of advanced statistical and machine learning methods is proposed, in combination with data reconciliation methods. Finally, for task (c), information acquired will be utilized to adapt the model feasible space by building accurate surrogate models and adaptively refine them using the online data acquisition. Although the focus will be on pharmaceutical production processes, the proposed work, if successful, can have significant broader impacts on a variety of industrial processes. Two PhD students will be trained on the development of a cutting-edge framework for autonomous manufacturing processes.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.