information in historical defect texts

 

 

MA Run ze, WANG Long xiang, YU Jia wen, WANG Hui fang, QIU Jian

 

(College of Electrical Engineering, Zhejiang University, Hangzhou 310027, China)

 

 

Abstract: Aiming at the feature that the condition evaluation of the circuit breaker(CB) is based on the operation information currently, while it does not involve CB′s historical defects which reflect the reliability of CB′s components, and the problem of the treatment and analysis of CBs′ Chinese defect texts, a comprehensive evaluation model of the CBs′ condition integrating the CBs′ operation information and historical defect information was proposed, which adds the evaluation rules for historical defect texts to the existing evaluation guidelines, and also a classification method of the defect texts was proposed based on KNN algorithm. The steps of the method is firstly the Chinese text preprocessing was done by doing word segmentation, word frequency statistics, deleting stop word and text vectorization, which were suitable for short texts, then the KNN algorithm was used to classify the defect texts. An example was used to verify the defect text classification method and the comprehensive evaluation model considering the information in historical defect texts. The results indicate that the classification method is feasible and the rationality of the CB′s evaluation results is improved.

 

Key words: circuit breaker; condition evaluation; defect texts; KNN algorithm