Graph Model for Probabilistic Resilience and Recovery Planning of Multi-Infrastructure Systems
Graph Model for Probabilistic Resilience and Recovery Planning of Multi-Infrastructure Systems
复制标题
DOI：
10.1061/(asce)is.1943-555x.0000338
复制
发表时间：
2017-09-01
影响因子：
3.3
通讯作者：
Hay, Alexander H.
中科院分区：
文献类型：
--
作者：
Bristow, David N.;Hay, Alexander H.
在全球范围内，冲击和压力的间接后果正在增加。评估和制定应对这些风险的计划仍然是一项挑战。困难在于，在通常只有高水平的时间统计数据可用时，提前确定损失和潜在的解决办法可能如何通过系统的复杂关联系统的相互作用传播。本文提出了一种新的图模型，结合最大熵似然估计器来映射复杂的相互作用，并评估不同启动场景所产生的统计量。该方法包括根据所有危害效应定义初始条件，然后从高级统计数据生成事件序列，以估计在冲击或压力后不同结果的概率。通过这种方式，设计了一种通用的方法来评估运营损失方面的间接后果，并评估风险处理方案的优点。将这些间接后果计算与直接后果评估结合起来，为以多目标方式平衡处理这些形式的后果提供了一条途径。
Indirect consequences to shocks and stresses are mounting globally. Estimating and developing plans to treat these risks remains a challenge. The difficulty is determining ahead of time how the losses, and the potential resolutions, might propagate through the interactions of complex connected systems of systems when generally only high-level temporal statistics are available. In this paper, a novel graph model coupled with a maximum entropy likelihood estimator are proposed to map the complex interactions and to assess the statistics resulting from different initiating scenarios. The method involves defining initial conditions from all-hazards effects, followed by generation of event sequences from the high-level statistics, to estimate probabilities of different outcomes after a shock or stress. In this way, a general method is devised to assess indirect consequence in operational loss terms and to assess the merits of risk treatment options. Incorporating these indirect consequence calculations with direct consequence assessments provides a way to support a balance of treatment of these forms of consequence in a multiobjective fashion.