A recursive Bayesian approach for fatigue damage prognosis: An experimental validation at the reliability component level
A recursive Bayesian approach for fatigue damage prognosis: An experimental validation at the reliability component level
复制标题
DOI：
10.1016/j.ymssp.2013.10.014
复制
发表时间：
2014-04
影响因子：
8.4
通讯作者：
M. Gobbato;J. Kosmatka;J. Conte
中科院分区：
文献类型：
--
作者：
M. Gobbato;J. Kosmatka;J. Conte
对于在使用寿命期间承受循环和随机载荷的各种机械和结构系统来说，疲劳引起的损坏是最不确定和高度不可预测的失效机制之一。因此，最终需要一个健康监测系统，能够（i）通过无损评估（NDE）技术监测这些系统的关键部件，（ii）评估其结构完整性，（iii）递归预测其剩余疲劳寿命（RFL），以及（iv）提供经济高效的基于可靠性的检查和维护计划（RBIM）。为了实现这些目标，本文的第一部分概述和扩展了基于可靠性的综合疲劳损伤预测方法（由作者之前开发），用于递归预测和更新航空航天结构中关键结构部件和/或子部件的 RFL。在本文的第二部分中，文献中提供的一组实验疲劳测试数据用于在可靠性组件级别（即在单个损伤位置演变的单一损伤机制）对所提出的框架进行数值验证和实验验证。本研究获得的结果证明了（i）近乎连续的 NDE 监测系统的重要性和好处，（ii）递归贝叶斯更新方案的效率，以及（iii）所提出的框架在递归更新和改进 RFL 估计方面的稳健性。这项研究还表明，所提出的方法可以导致一定程度的 RFL（从而在不影响最低安全要求的情况下获得经济收益）或通过检测过早故障来提高安全性，从而避免代价高昂的灾难性故障。
Fatigue-induced damage is one of the most uncertain and highly unpredictable failure mechanisms for a large variety of mechanical and structural systems subjected to cyclic and random loads during their service life. A health monitoring system capable of (i) monitoring the critical components of these systems through non-destructive evaluation (NDE) techniques, (ii) assessing their structural integrity, (iii) recursively predicting their remaining fatigue life (RFL), and (iv) providing a cost-efficient reliability-based inspection and maintenance plan (RBIM) is therefore ultimately needed. In contribution to these objectives, the first part of the paper provides an overview and extension of a comprehensive reliability-based fatigue damage prognosis methodology — previously developed by the authors — for recursively predicting and updating the RFL of critical structural components and/or sub-components in aerospace structures. In the second part of the paper, a set of experimental fatigue test data, available in the literature, is used to provide a numerical verification and an experimental validation of the proposed framework at the reliability component level (i.e., single damage mechanism evolving at a single damage location). The results obtained from this study demonstrate (i) the importance and the benefits of a nearly continuous NDE monitoring system, (ii) the efficiency of the recursive Bayesian updating scheme, and (iii) the robustness of the proposed framework in recursively updating and improving the RFL estimations. This study also demonstrates that the proposed methodology can lead to either an extent of the RFL (with a consequent economical gain without compromising the minimum safety requirements) or an increase of safety by detecting a premature fault and therefore avoiding a very costly catastrophic failure.