A quality evaluation method for the unstructured defect record of relay protection devices based on ontology and knowledge graph

Abstract

The original unstructured record data for the defect of the relay protection devices (RPDs) may contain problems influencing the data mining, and it is lack of quantitative evaluation. So the purpose of this paper is to evaluate the quality of these defective texts of relay protection devices, and it proposes a text quality evaluation method for the unstructured defect record of RPDs based on the ontology and knowledge graph for the defects text of relay protection devices, combined with the analytic hierarchy process (AHP) and entropy weighting method. In details, first, the problems existing in the unstructured defect records of RPDs are presented. Secondly, the corresponding evaluation indicators are proposed, and the quantitative calculation method for each indicator is given in combination with the knowledge graph for the defects of relay protection devices. Thirdly, combined with the subjective AHP and the objective entropy weight method, a comprehensive evaluation method for the unstructured defect record for RPDs is proposed, and its effectiveness is verified with examples. Finally, with the 10-year actual defects record data, the data quality of the unstructured defect record for the RPDs in a regional power grid is comprehensively evaluated, and characteristic of the data quality with respected to text length, defect level, and dispatch region are revealed.

Keywords

• Relay protection device
• Defect record
• Knowledge graph
• AHP-entropy weight method
• Text quality evaluation