Network Target Coordination for Complex System Optimization

该奖项的研究目标是开发一种基于优化的设计方法，该方法考虑了复杂系统的动态特性。该方法将具有建模能力，从概念上将对象分解（通过物理子系统）与方面分解（通过规程）集成到节点网络中，在该网络中，根据来自其他节点的性能目标执行局部优化。整体网络优化将被协调，以实现设计性能与子系统和学科性能之间的理想折衷。双重分解方法将使方法全面并响应不断变化的设计条件，而基于代理的实现将保证计算灵活性。该方法将在过程中纳入人类决策，并将成为在布局设计问题上实施的工具的基础。如果成功，设计师将能够为复杂系统（如交通工具或飞机）做出决策，同时考虑到整个变化条件范围的失败和机会。这项研究的成果将透过各种渠道传播，包括工程和数学科学的研究生培训、跨学科研讨会、专业会议、出版物、开放源码软件和互联网网站。该方法将提交给陆军（TACOM）、其他联邦机构（空军、能源部、NASA、海军）和原始设备制造商公司（宝马、福特、通用、波音）。研究生将参与部分拟议的研究，而这项工作的结果将包括在研究生水平的课程。拟议的工作将提供机会，培养学生谁可以进入劳动力与先进的数学技能，并了解如何将这些应用于复杂的工程问题。将作出特别努力，向有很大比例代表性不足人口的地方学校的学生伸出援助之手。

The research objective of this award is to develop an optimization-based design methodology that accounts for the dynamic nature of complex systems. The methodology will have a modeling capability to conceptually integrate object decomposition (by physical subsystems) with aspect decomposition (by disciplines) into a network of nodes at which local optimizations are performed subject to the performance targets from the other nodes. The overall network optimization will be coordinated to accomplish a desired compromise between the performance of the design and the performance of the subsystems and disciplines. The dual-decomposition approach will make the methodology comprehensive and responsive to changing design conditions while an agent-based implementation will guarantee computational flexibility. The methodology will incorporate human decision-making in the process and will be the basis for a tool implemented on a layout design problem. If successful, designers will be able to make decisions for complex systems (such as vehicles or airplanes), while accounting for failures and opportunities resulting from the consideration of the entire spectrum of changing conditions. The outcomes of this research will be disseminated through various channels including graduate student training in engineering and mathematical sciences, interdisciplinary seminars, professional meetings, publications, open source software, and Internet websites. The methodology will be presented to the Army (TACOM), other federal institutions (Air Force, DOE, NASA, Navy), and original equipment manufacturer companies (BMW, Ford, GM, Boeing). Graduate students will be involved in parts of the proposed research while results of this work will be included in graduate-level courses. The proposed work will provide opportunities to train students who can enter the workforce with advanced mathematical skills and an understanding of how to apply these to complex engineering problems. Special efforts will be undertaken to reach out to students from local-area schools that have a significant proportion of underrepresented populations.