SDG fault diagnosis based on Granular Computing and its application

Signed Directed Graph (SDG) fault diagnosis method can be used to express complicated cause-effect relationship, and has the capacity of containing large-scale potential information, it is a self-contained method to effectively diagnose system failures, but SDG model contains redundant information, increasing the computational complexity, and diagnoses lists more relevant results, resulting in low-resolution. In order to solve these problems, the attribute reduction algorithm based on Granular Computing (GrC) is introduced in to remove redundant attributes and identify the minimal attribute reduction, and then, granule is used to formally express the elements of the decision table, after that the granular base of decision-making rules is constructed, granule reasoning method is used to obtain the most possible fault source by computing the most similarity. Finally, the power plant deaerator is taken as an example, which illustrates this method is valid.