Diagnostic Bayesian networks modeling based on multi-signal flow graphs

The similarities and differences between multi-signal flow graphs and Bayesian networks were analysed, which show that the conversion of multi-signal flow graphs to Bayesian networks is feasible. A diagnostic Bayesian networks modeling method was proposed based on multi-signal flow graphs in order to reduce the diagnostic Bayesian network models development cost. The method includes the conversion of the structure of multi-signal flow graphs to the structure of Bayesian networks and the conversion of information of multi-signal flow graphs to the conditional probabilities table of Bayesian networks. The diagnostic Bayesian network model is good at dealing with uncertainty and conflicting evidence in complex systems fault diagnosis. The experiment results show that the diagnosis accuracy of diagnostic Bayesian network models is the same with multi-signal flow graph models.