Algorithms for optimal risk-based planning of inspections using influence diagrams

Algorithms for optimal risk-based planning of inspections using influence diagrams
复制标题

使用影响图进行基于风险的最佳检查计划的算法

DOI：
--
复制
发表时间：
2013
期刊：
影响因子：
--
通讯作者：
D. Štraub
D. Štraub
中科院分区：
--
文献类型：
--
作者：
L. Luque;D. Štraub

文献摘要

被引文献

相似文献

研究了基于风险的影响图检验优化方法。 为此，使用动态贝叶斯网络(DBN)方法建立了疲劳劣化模型 提出了一种新的解决方案。DBN包含了以前检查活动中的信息。 在网络内部定义决策和公用事业节点以表示检查 和维修活动。最优检查策略(受安全或公用设施约束) 使用有限记忆影响图(LIMID)方法进行近似， 并采用单策略更新、局部优化策略进行求解。在一个 数值研究发现，这种方法给出的解略好于 用以前应用的简单启发式方法获得的结果，如可靠性 阈值或定期检查启发式。最后，通过数值算例说明了 自适应检查战略的优势，即根据 以前的检查结果。
Risk-based optimization of inspection using influence diagrams is investigated. To this end, a fatigue deterioration model using a Dynamic Bayesian Network (DBN) approach is presented. The DBN incorporates information from previous inspection campaigns. Decision and utility nodes are defined inside the network to represent inspection and repair activities. The optimal inspection strategy (subject to safety or utility constraints) is approximated using the Limited Memory Influence Diagram (LIMID) approach, and is solved using the single policy updating, a local optimization strategy. In a numerical investigation, this method is found to give solutions that are slightly better than those obtained with simple heuristics that were previously applied, such the reliability threshold or periodic inspection heuristic. Finally, the numerical example demonstrates the superiority of adaptive inspection strategies, whereby inspections are planned based on the results of previous inspections.