Development of Rule-based Diagnosis for the Automatic Diagnosis of Vertical Pumps

Recent trends in facility maintenance have focused on automating defect diagnosis using artificial intelligence(AI) technologies. One such method utilizing AI is rule-based diagnosis, which operates based on decisions made from expert knowledge and does not require large datasets. However, because it evaluates only the data measured at a single point, the scope of evaluation is limited and requires further development. This paper presents the development of a multi-point rule-based diagnostic system that evaluates the entire system by considering measurement results from multiple points. The performance of the proposed diagnostic method is validated by comparing the results of direct diagnoses with those obtained through rule-based diagnosis, using durability test data from a vertical pump. The durability test results were categorized into four conditions based on their progression, and the diagnostic outcomes were compared at each step.