系统实时性能可靠性评估的贝叶斯方法

多数产品或系统的失效可归结为其潜在的退化过程，且系统退化通常为隐含的过程而无法直接观测或测量。针对隐含的退化过程不可直接观测的特点，本文提出一种基于状态空间模型的产品退化建模方法，以系统运行期间采集的性能数据为基础，结合贝叶斯估计理论，实现了对系统隐含退化状态的在线估计与预测，以及对系统性能可靠性的实时评估。以民航发动机为对象，通过对实际收集的性能数据的分析表明，本文所提方法能够较好地用于系统性能可靠性的实时评估。

Bayesian method for system real-time performance reliability assessment

The failure mechanisms of many products can be attributed to an underlying degradation, and typically the latent degradation state cannot be measured directly. For that feature a state-space model is proposed to describe the degradation process. Given the performance data collected from the in-service system, the latent degradation state can be estimated and its future projection is predicted in the Bayesian framework, then the performance reliability is evaluated. The developed method is applied to civil aircraft engines for real-time performance reliability assessment, and the results presented in the case study show that the developed method can work well for system real-time performance reliability estimation.