层次因子图及其在故障诊断中的应用

Hierarchical factor graph and its application in fault diagnosis

为解决大规模复杂不确定性故障诊断问题的建模和推理难题,提出一种扩展的因子图模型——层次因子图,并给出了相应的推理及建模方法。该模型采用层次式拓扑结构表达问题领域的结构化知识,具备现有概率图模型的所有优点。经实例验证表明:所提出的层次因子图模型可以较好地解决大规模复杂不确定性故障诊断问题。

An extended factor graph model named hierarchical factor graph （HFG）, along with infe- rence and modeling methods, is proposed to solve a lot of difficult and complicated problems of mode- ling and inference in fault diagnosis. Hierarchical structure is used in HFG to represent structural knowledge in the domains of application with all merits of the existing probabilistic graphical models. The experiment results show that the proposed HFG can effectively solve a large number of complica- ted and uncertain problems in fault diagosis.