The Extraction of Hidden Fault Diagnostic Knowledge in Equipment Technology Manual Based on Semantic Annotation

Abstract

Due to small quantities, lack of service experience, and poor fault diagnosis knowledge of new-type equipment, it is often difficult to determine the exact location of a trouble. To address this problem, a knowledge capitalization and fault diagnosis method based on semantic annotation was proposed, which can extract deep fault knowledge implied in the technical publications. Firstly, the unstructured nature of deep fault knowledge in the technical publications is outlined. And the role of semantic annotation in the process of knowledge acquisition is highlighted. Secondly, an ontology model for deep fault diagnosis knowledge extraction is developed to annotate the technical publications semantically. And the annotation method is presented to translate the unstructured and implicit knowledge into formal-defined and computer readable semantic net. Then, a fault diagnostic algorithm is proposed to use the annotation results based on hierarchical diagnosis algorithm of directed graph. Finally, an application case of VE-type fuel-injection pump verifies the feasibility and effectiveness of this method.