Modeling Uncertain and Dynamic Interdependencies of Infrastructure Systems Using Stochastic Block Models
Modeling Uncertain and Dynamic Interdependencies of Infrastructure Systems Using Stochastic Block Models
复制标题
DOI：
10.1115/1.4046472
复制
发表时间：
2020-06
期刊：
影响因子：
--
通讯作者：
Jin-Zhu Yu;H. Baroud
中科院分区：
文献类型：
--
作者：
Jin-Zhu Yu;H. Baroud
对相互依赖的关键基础设施（ICI）的弹性进行建模需要仔细评估这些系统的相互依赖性，因为这些系统变得越来越相互关联。由于缺乏相关数据，ici之间的相互依赖联系往往存在不确定性。然而，这种不确定性还没有得到适当的描述。本文提出了一种基于概率图模型的集成电路弹性建模方法。采用随机块模型（sbm）对集成电路之间相互依赖联系的不确定性进行建模。具体来说，该方法估计在SBM中被视为块的单个系统之间的链接的概率。提出的模型采用几个属性作为预测因子。提出并比较了基于静态和动态分量重要性排序的两种恢复策略。该方法以田纳西州谢尔比县相互依赖的水电网络为例进行了说明。结果表明，相互依赖联系的概率因估计中考虑的预测因子而异。考虑到相互依赖环节中的不确定性，可以实现动态恢复过程。基于组件重要性排序动态更新的恢复策略加快了恢复速度，从而提高了集成电路的弹性。
Modeling the resilience of interdependent critical infrastructure (ICI) requires a careful assessment of interdependencies as these systems are becoming increasingly interconnected. The interdependent connections across ICIs are often subject to uncertainty due to the lack of relevant data. Yet, this uncertainty has not been properly characterized. This paper develops an approach to model the resilience of ICIs founded in probabilistic graphical models. The uncertainty of interdependency links between ICIs is modeled using stochastic block models (SBMs). Specifically, the approach estimates the probability of links between individual systems considered as blocks in the SBM. The proposed model employs several attributes as predictors. Two recovery strategies based on static and dynamic component importance ranking are developed and compared. The proposed approach is illustrated with a case study of the interdependent water and power networks in Shelby County, TN. Results show that the probability of interdependency links varies depending on the predictors considered in the estimation. Accounting for the uncertainty in interdependency links allows for a dynamic recovery process. A recovery strategy based on dynamically updated component importance ranking accelerates recovery, thereby improving the resilience of ICIs.