A Data-Knowledge Hybrid Driven Method for Gas Turbine Gas Path Diagnosis

Gas path fault diagnosis of a gas turbine is complex task involving field data analysis and knowledge-based reasoning. In this paper, data-knowledge hybrid driven method for proposed by integrating physical model-based (GPA) with ontology model. Firstly, GPA used to extract the features from data. Secondly, virtual distance mapping algorithm developed map result specific feature criteria individual described in Finally, model built support automatic reasoning maintenance strategy mapped individual. To enhance ability selecting proper strategy, represents more abundant knowledge several sources, such as analysis, structure FMECA (failure mode, effects, criticality analysis), logic decision tool. The availability verified real GE LM2500 PLUS unit. results indicate that effective detecting advance. Furthermore, diversified information, criticality, consequence, detectability, could be provided strategy. It proven can improve capability detection, selection.