Optimal inspection of binary systems via value of information analysis
Optimal inspection of binary systems via value of information analysis
复制标题
通过信息分析的价值对二元系统进行优化检查
DOI：
10.1016/j.ress.2021.107944
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
Chaochao Lin, Junho Song
中科院分区：
文献类型：
--
作者：
Chaochao Lin, Junho Song
我们开发了可计算的指标来分配优先级的二进制系统组成的二进制组件的信息收集。组件值得检查，因为它们的条件状态是不确定的，系统功能依赖于它们。信息价值（VoI）可以评估信息在不确定性情况下对决策的影响，包括组件的可靠性和在系统中的作用，观测的精度，可用的维护措施和预期的经济损失。我们介绍了基于VoI的系统级（“全球”）和组件级（“本地”）的维护行动的指标，分析这些指标的属性，并将其应用到串联和并联系统。我们讨论了它们的计算复杂性，在一般的网络系统中的应用，驯服的复杂性，局部度量评估，我们提出了一个启发式和评估其性能的一些案例研究。
We develop computable metrics to assign priorities for information collection on binary systems composed of binary components. Components are worth inspecting because their condition states are uncertain, and system functioning depends on them. The Value of Information (VoI) enables assessment of the impact of information in decision making under uncertainty, including the component’s reliability and role in the system, the precision of the observation, the available maintenance actions and the expected economic loss. We introduce the VoI-based metrics for system-level (“global”) and component-level (“local”) maintenance actions, analyze the properties of these metrics, and apply them to series and parallel systems. We discuss their computational complexity in applications to general network systems and, to tame the complexity for the local metric assessment, we present a heuristic and assess its performance on some case studies.