基于贝叶斯网络的复杂系统故障诊断

系统结构和部件关系复杂、试验费用昂贵是小样本下基于不确定性信息的决策问题。针对其特点，建立了基于贝叶斯网络的复杂系统故障诊断模型，并提出采用Leaky Noisy-OR模型来降低数据需求量和计算复杂度。经研究表明，这种方法能综合利用各种来源信息，具有知识表达明确、样本需求量小、故障诊断准确度高等特点，可为复杂系统故障诊断提供决策支持。

Fault Diagnosis of Complex System Based on Bayesian Networks

The structure and relationship of components are complicated in complex system, and the test-cost of complex system is high. So, the fault diagnosis of complex system is the decision with uncertainty under small sample. A Bayesian networks (BN) model for fault diagnosis of complex system is built up,furthermore, the Leaky Noisy-OR model is used to reduce the requirements for data and computation complexity. The study shows that the method for fault diagnosis of complex system can utilize all kinds of information, represent knowledge definitely, demand small sample and diagnose the faults accurately. The method can guide the decision for fault diagnosis of complex system.