Collaborative Research: An Integrated Approach to Modeling, Decision-Making and Control for Energy Efficient Manufacturing

该项目将支持以提高能源效率和促进制造业健康室内环境为重点的基础研究。为了实现美国积极的脱碳目标，除了当前的能源生产结构外，还需要在需求侧能源管理方面进行重大转变。在典型的制造设施中，最重要的能源消耗来源是制造系统和环境控制系统，如供暖、通风和空调。这两个系统在生产操作、动态能源需求/消耗和室内条件方面紧密相连。然而，目前制造系统的控制策略缺乏与设施能源管理和室内环境控制的有效整合，阻碍了整体效率的提高。该基金支持多学科研究，以建立对智能制造设施能源效率的全面了解，以减少能源浪费，提高整体制造效率，降低制造成本，并促进工业工人的福祉。这项研究的成果将为环境、社会和美国能源格局带来长期效益。此外，这项研究符合工业需求，促进多样性，鼓励代表性不足的群体参与研究，并为工程教育的进步做出贡献。本研究致力于开发复杂系统集成建模、多智能体决策和分布式控制的创新技术。研究团队的目标是建立制造系统和环境控制系统的动态模型，以全面了解制造设施中各个组件之间的动态相互作用。此外，将使用图神经网络建立一个集成的工厂能源模型，弥合传统上分离的环境控制管理和制造系统之间的差距。此外，考虑到生产操作和设施能源管理，将设计一个分层控制框架，以整合监督决策和自适应控制方案。该团队将开发一种多智能体强化学习算法，以支持图神经网络描述的复杂系统的在线决策。数据驱动的自适应控制算法将使用基于学习的方法来处理系统的不确定性和环境干扰。这项基础研究有可能克服传统稳态分析应用于分离制造系统和环境控制的局限性，将能源和生产效率提升到新的水平。此外，这些方法的通用性将有助于更广泛的工程系统建模和控制领域。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 project will support fundamental research focused on improving energy efficiency and promoting a healthy indoor environment in the manufacturing industry. To achieve the United States' aggressive decarbonization goal, a significant transformation in demand-side energy management is needed alongside the current energy generation mix. In typical manufacturing facilities, the most significant sources of energy consumption are the manufacturing systems and the environmental control systems, such as heating, ventilation, and air-conditioning. These two systems are closely interconnected in terms of production operations, dynamic energy demand/consumption, and indoor conditions. However, the current control strategy for manufacturing systems lacks effective integration with facility energy management and indoor environmental control, hindering overall efficiency improvements. This grant supports multi-disciplinary research to establish a comprehensive understanding of energy efficiency in smart manufacturing facilities to reduce energy waste, enhance overall manufacturing efficiency, lower manufacturing costs, and promote the well-being of industry workers. The outcomes of this research will yield long-term benefits for the environment, society, and the U.S. energy landscape. Moreover, this research aligns with industrial needs, fosters diversity, encourages the involvement of underrepresented groups in research, and contributes to the advancement of engineering education.This research endeavors to develop innovative technologies for integrated modeling of complex systems, multi-agent decision-making, and distributed control. The research team aims to construct dynamic models of manufacturing systems and environmental control systems, gaining comprehensive insights into the dynamic interactions among various components of the manufacturing facility. Furthermore, an integrated factory energy model will be established using a graph neural network, bridging the gap between traditionally separate management of environmental control and manufacturing systems. Additionally, a hierarchical control framework will be designed to integrate supervisory decision-making and adaptive control schemes, considering both production operations and facility energy management. The team will develop a multi-agent reinforcement learning algorithm to support online decision-making for the complex system described by the graph neural network. Data-driven adaptive control algorithms will be employed to handle system uncertainties and ambient disturbances using a learning-based approach. This fundamental research has the potential to overcome the limitations of traditional steady-state analysis applied to separate manufacturing systems and environmental control, elevating energy and production efficiency to new levels. Furthermore, the generic nature of the methods will contribute to the broader field of engineering system modeling and control.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.