A Method of Fuzzy Reasoning Based on Semantic Similarity and Bipartite Graph Matching

For the shortcomings of fuzzy rule reasoning method in traditional fault diagnosis expert system, a fuzzy rule matching method based on semantic similarity and bipartite graph is proposed in this paper. Firstly, fault domain knowledge ontology is built by analyzing domain fault knowledge, and then the input fault phenomena set and the property set of rule set antecedent act as the vertex sets of bipartite graph. The edge sets are made up of the lines between properties matched, and the weight of edges are semantic similarity between nodes, thus the problem becomes the optimal matching of bipartite graph. The experiment shows that the method can improve the efficiency of rule matching, which offers a feasible solution for fault diagnosis of complex system.