基于贝叶斯最大后验概率准则的大型复杂系统故障诊断方法研究

A Fault Diagnosis Approach of the Large Complex System Based on Bayes Theory

针对大型复杂系统的特点,在故障空间建立故障源—测试依赖关系模型。在此基础上,将贝叶斯理论引入大型复杂系统故障诊断,以最大后验故障概率为准则判断系统所处的运行状态。将该问题归纳为集合覆盖问题(SCP),用拉格朗日松弛算法求得近优解。将算法应用到阿波罗飞船发射前故障诊断实例中,取得了满意的实验结果。

According to the characteristic of the large complex system, we built a fault diagnosis model in fault space to describe the relationship between fault sources and test set. Based on it, Bayes theory was introduced into the problem of fault diagnosis of the large system. The running states of the system were estimated according to the maximal posterior probability. We concluded the problem as a set cover problem （SCP）, and the near-optimal resolution was given by Lagrange relax algorithm. The example in which the algorithm is applied to the fault diagnosis of the Apollo prelaunch data guoted from Ref. [11] demonstrates the algorithm has enough precision and efficiency.