贝叶斯网络故障诊断模型解耦方法的探讨

With too may nodes and complicated coupling relationship, complex system bayesian network fault diagnosis model has low reasoning diagnosis efficiency and bad explanation ability for the diagnosis results. Using decoupling thought, through working on the three typical coupling sub-BNs (The coupling node points to sub-BNs, the indirectly coupling non-symmetrical sub-BNs, the "V"coupling sub-BNs )decoupling, and then compare the reasoning diagnosis results before and after decoupling networks, verified the feasibility of decoupling thought, so as to improve the bayesian network fault diagnosis model diagnosis efficiency. This paper actually based on the research of the fault diagnosis BN model in train, through those three decoupling experiments, to research the decoupling method. Finally, the conclusion prove the efficiency of the method.