多变量条件下状态维修建模理论研究

基于状态的维修理论是可修系统可靠性研究的前沿课题。大多数现有的模型和方法考虑单个状态变量或者多个状态变量的线性组合的情况，并假设其失效门槛是已知的、固定不变的。这些与大多数工程实际问题不符。本项目拟突破这些限制。项目将考虑多个状态变量，以更符合实际的方式将它们融合为一个复合状态变量；视复合状态变量为一个随机变量，其失效门槛不再是预先给定的值，而是服从某种分布，并基于观察的信息来确定。在这种背景下，主要的研究课题为：1）多变量非线性融合技术研究；2）复合状态变量的失效门槛建模技术研究；3）复合状态变量的警报门槛建模技术研究；4）基于状态信息的剩余寿命分布的推断技术研究。

Condition-based maintenance theory is a research frontier in repairable system reliability. Most of the existing models and methods consider a single condition variable or a linear combination of several condition variables; the failure threshold is assumed to be known and fixed. These are inconsistent with most of practical problems. The proposed research proposal aims to break through these limitations. The project will consider the case where there are multiple condition variables, which will be fused into a composite condition variable in a way more consistent with practices. The composite condition variable is viewed as a random variable so that the failure threshold is no longer pre-specified; instead, it follows a certain distribution and will be determined based on the observed information. In this setting, the main research topics are: 1) research on multivariate nonlinear fusion techniques; 2) research on the failure threshold; 3) research on the alarm threshold; and 4) research on condition-information-based inference techniques for residual life distribution.