1.College of Electrical Engineering, Zhejiang University, Hangzhou310027, China;2.Zhejiang Huayun Information Technology Co., Ltd., Hangzhou310012, China
Power equipment defect text contains a lot of historical defect information of the equipment. The accurate information identification from defect text can provide effective guidance for equipment failure rate modeling and health status evaluation. By introducing dependency parsing technology, “left-greedy” out stack rule and neural network based dependency state transition analysis model are proposed. This paper implements the dependency-syntax-tree construction of actual defect text and defect classification standard text of power equipment, and proposes a method of pruning, segmentation and reconstruction of power equipment dependency syntax tree using the characteristics of the defect text. Meanwhile, dependency relationship based tree matching algorithm is proposed to match actual defects with standard defect dependency syntax trees. The feasibility and effectiveness of defect information identification method based on dependency syntax analysis is illustrated with the example of main transformer defect text. The results show that the efficiency and accuracy of the proposed method are improved substantially compared with other tree matching algorithms and semantic similarity computation methods.
|||SHAO Guanyu, WANG Huifang, WU Xianghong, et al. Precise Information Identification Method of Power Equipment Defect Text Based on Dependency Parsing[J]. Automation of Electric Power Systems,2020,44(12):178-185. DOI:10.7500/AEPS20190401001|