Fault diagnosis method based on probability extended SDG and fault index

摘要

A novel fault diagnosis method based on probability extended SDG (Signed Directed Graph) and fault index inference has been proposed. First, the SDG model is extended by fuzziflcation method and probability distribution. Furthermore, the model structure is simplified to reduce the search space, and the set of fault source candidates is obtained by the hierarchical method. When the system has multiple fault candidates, the fault source can be identified according to the rank order of fault index, which is calculated by fuzzy deviation, fault propagation speed and marginal probability without path information. The reactor of Tennessee Eastman process (TEP) has been taken as a simulation example. The result shows that the diagnosis method has high fault resolution and is suitable for fault location.