Automated Methods for Modeling and Designing Resilient Complex Networks

网络出现在许多工程环境中，包括制造、国防、能源和运输。对部分网络的干扰可能导致性能显著下降或功能完全中断。该项目将提高我们建模和设计网络的能力，这些网络在受到各种干扰时，能够在指定的运营预算内保持足够的性能和功能。这项研究的结果将在供应链网络的背景下得到证明，供应链网络是高度互联的结构，随着企业交换商品以创造最终产品而出现。确保供应链功能对工业竞争力和国防至关重要，但许多现有网络都是基于过度简化的模型，容易受到干扰和性能下降的影响。这项研究将通过为复杂场景提供更明智的网络设计决策来改善最先进的技术，并通过改进供应链的设计来证明这一点。这项工作的成果将包括一般的数学技术和具体的设计准则。该项目将在这一至关重要的领域对学生进行培训，并编写新的教材。实现项目目标需要开发有效的算法，并在网络科学和博弈论方面取得进展。为了表示复杂网络，将利用基于动作的系统模型并专门用于供应链环境。模型参数将使用从公开的真实世界供应网络获得的多层复杂网络输入和由与供应链网络相关的目标指导的多目标优化算法来找到。充分和完整地表示供应链网络的能力将允许需要零模型的测试或检查假设场景。第二个项目的目标是设计高效的自动机制设计算法，为企业设计激励措施，以创建连接，从本质上确保网络的鲁棒性扰动。这些被称为复杂的网络形成游戏，理论将推导出形式化。第三个项目目标利用建模框架进行集中设计，而不是对特定网络进行建模，并将从分散博弈中获得的结果进行比较-该奖项反映了NSF的法定使命，并被认为值得通过利用基金会的智力价值和更广泛的影响审查标准。

Networks arise in numerous engineering contexts, including manufacturing, defense, energy, and transportation. Disturbances to part of a network can result in significant performance degradation or complete disruptions of functionality. This project will advance our ability to model and design networks capable of maintaining sufficient performance and functionality within a specified operational budget when subjected to various types of perturbations. Results of this research will be demonstrated in the context of supply chain networks, which are highly interconnected structures that arise as firms exchange goods to create final products. Ensuring supply chain functionality is critically important to industrial competitiveness and national defense, but many existing networks are based on over-simplified models that are susceptible to disruption and performance degradation. This research will improve the state of the art by enabling better-informed network design decisions for complex scenarios and demonstrate this through the design of improved supply chains. Outcomes of this work will include general mathematical techniques and specific design guidelines. The project will train students in this area of critical importance and produce new educational material. Accomplishing the project goals requires the development of efficient algorithms and advances in network science and game theory. In order to represent complex networks, an action-based system model will be utilized and specialized to the supply chain context. Model parameters will be found using multi-layer complex network input obtained from publicly available real-world supply networks and a multi-objective optimization algorithm that is guided by objectives relevant to supply chain networks. The ability to adequately and compactly represent a supply chain network will allow for tests requiring a null-model or to examine what-if scenarios. The second project goal is to devise efficient automated mechanism design algorithms for devising incentives for firms to create connections that inherently ensure network robustness to perturbations. These are termed complex network formation games, and theory will be derived to formalize them. The third project goal utilizes the modeling framework for centralized design, instead of modeling a specific network, and will compare the results obtained from the decentralized game- theory perspective in order to ascertain differences between the quality of resultant networks and to extract best design principles.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.