Model and Fault Inference with the Framework of Probabilistic SDG

As the scale of systems increases, traditional models and fault diagnosis methods are not applicable. Qualitative signed directed graphs (QSDG) are used to model the variables and relationships among them in large-scale complex systems. However, they have distinct limitations of resulting spurious solutions due to the lack of utilization of knowledge or information. This article proposes a kind of probabilistic SDG (PSDG) model to describe the propagation of faults among variables. The fault diagnosis method is also investigated, where Bayesian network has been employed. Finally, examples are given and the future topics are listed